P2PContent

Definitions
Backbone: The Internet backbone is the part of the Internet that connects Internet Service
Providers and other organizations with large to very large networks to each other. Some
providers of this type of service are UUNET, Williams Communications and Level 3
Communications.
Bandwidth: The amount of bits per second to which a network connection is limited. A T1
circuit, for example, carries 1.544 megabits per second, so 1.544 Mbps is its bandwidth.
Typical dialup users are connecting at anywhere from 300 bits per second -- 56 kilobits per
second. However, it's unlikely that a user who is connecting at much less than 33.6 Kbps will
have the ability to use any form of audio or video content distribution. Typical home broadband
connections might run anywhere from 100Kpbs - 1.5 Mbps.
Generally speaking, a T-1 is too small to be of any practical use to a major content provider, or
to a P2P service provider unless the intent is test usage only. These data rates are the ones most
likely to come up in major content server or discussions.
T-3 45 Mbps
OC-3 155 Mbps
OC-12 622.08 Mbps
OC-24 1.244 Gbps
OC-48 2.488 Gbps
OC-192 10 Gbps
Alternatively, a content provider might use one of the service providers such as Yipes that
provide direct 100 Mbps Ethernet connections to the backbones to which they are connected.
Content delivery network (CDN): A network of servers and intermediate connections,
either owned or leased, designed to bring delivery of Internet content as quickly and smoothly
as possible to end users. Examples include Akamai, Digital Island, iBeam and Speedera.
Digital subscriber line: DSL uses a modified standard voice-type telephone line to deliver a
high-speed digital data connection considerably faster than dialup modems, using a specialized
DSL modem connected to the user's computer or router. It is frequently used to support home
and small business LANs and even a few Web servers. The data rate can be anywhere from
128k - 6.1 Mbps. It is usually run over the same wire pair that supports the user's voice
telephone connection.
A DSL connection may be symmetrical, i.e. the same speed in most directions (SDSL) or
asymmetrical (ADSL) meaning that it supports different data rates for incoming and outgoing

data. ADSL generally supports a fast incoming connection and a slower outgoing connection,
reflecting the fact that most end users use DSL for faster Websurfing and to get incoming
multimedia content. The majority of service providers provide ADSL service. At this point, the
telcos are the main source for DSL service, given the disappearance of independent DSL
service providers.
Distributed computing: In some contexts, this term can be used to describe software which
runs in a distributed manner across a client-server network. In this case, however, we refer to
the use of multiple computers, connected either to a company’s internal network or to the
Internet, linked together to share bandwidth, connectivity, or network-attached storage. See
also P2P.
Hosting company: A third-party provider who provides server access and network
connectivity to make customer data or other content available over the Internet.
ISP (Internet Service Provider): A company which provides connections to the Internet for
a fee. In recent times, the consumer Internet market has segmented into two types of access: 1)
broadband access, via cable modem,DSL (Digital Subscriber Line service), satellite or wireless
802.11b, and 2) traditional dial-up connectivity. See Bandwidth above for discussion of
typical bandwidth available to end users.
Latency: In the context we’re using the term, latency is the lag time between the moment an
end user requests Web information and the moment he or she gets the first packet of the content
they requested.
NSP (Network Service Provider): A company providing access to network facilities. Such
companies often provide Internet access but may also restrict themselves to offering private
connectivity for enterprises.
Peer to peer networking: In theory, P2P architectures allow communication and data sharing
between computing devices (including PCs, PDAs, cell phones and laptops) without the
intervention of any form of central server to handle either applications or file management. In
practice, most P2P architectures are “hybrid,” melding some form of server into their
functionality, but also allow device-to-device connectivity as well. This allows the injection of
customer content into one end of the network.
In the case of P2P content delivery networks, content is mediated via a central server. Not only
does the server provide the original content, but it also provides DRM (Digital Rights
Management), monitors the end points of the network to determine the optimum connection
paths between users, decides who is eligible to receive specific content, etc. In P2P CDNs,
once the content leaves the server it is eventually redistributed from user PC to user PC.

Quality of Service (QoS): A term referring, in most environments, to the ability of a network
service provider to provide a specified level of performance, often in terms of factors such as
data throughput and network uptime.
Streaming media: This term refers to the continuous transmission of audio or video data
across a network, typically but not necessarily the Internet.
Webcasting: Using a network such as the Internet to broadcast traditional media, e.g. audio
or video.

Quick Takeaway: 10 Key Points
1. Even If P2P CDNs Don’t Deliver On All of Their Claims, Their Cost Advantages
Will Shift The Marketplace.
In our research for this report, we’ve heard some contend that even if P2P players offer large
cost savings, it won’t matter, because they supposedly can’t offer Quality of Service guarantees
along the lines of a standard network service provider.
Well, we disagree. We believe the P2P streaming players’ quality and cost savings claims have
at least some merit, and more importantly, that their message is getting through to buyers in the
marketplace. In particular, we believe Webcasting customers , ISPs and telcos/backbone
providers are like to see P2P streaming as promising -- or at least worth testing.
As a result, non-P2P infrastructure players such as Akamai will find themselves will find
themselves under unprecedented downward price pressure, in all probability changing the way
they price and bundle their services.
2. The Consumer P2P Streaming Market Is Hot, But Enterprise Delivery Is A Bigger
Opportunity
Looked at one way, it’s a no-brainer. There are millions of Madonna fans eager to hear her
next single, and comparatively few enterprise users eager to review their company’s HR
orientation video.
On the other hand, enterprises have a well-defined need for affordable, well-managed
communication. And P2P streaming, which must be managed minutely to offer any sort of
quality, seems to fit the bill. Enterprise needs, ultimately, are where the stable source of bucks
lies.
3. Network Service Providers And Telcos Are The Ultimate P2P Streaming Channel
Since they’re still only toddlers, peer to peer content delivery companies are still making their
pitches in the marketplace largely on their own. But a few have already begun to do what we
think more or less all will do -- seek partnerships with ISPs and telcos to bring networking
muscle to their proposals.
4. The Logical Corollary: Hosting Companies Should Consider Partnering With,
Investing In or Acquiring Distributed Computing Companies to Offer Their Own
Streaming P2P Services

ISPs and Web hosting firms, let’s face it. If your customers do any sort of broadcasting, it’s
already better to join the P2P streaming companies than try to beat them. With estimated
bandwidth savings ranging from 50% to 90%, hosting firms and ISPs owe it to themselves to at
least give P2P streaming a try.
5. “Traditional” Content Delivery Networks Won’t Survive The Next Wave of
Consolidation And Competition Without an Infrastructure Partner
It’s no coincidence that Williams owns 49% of iBeam and Cable & Wireless picked up all of
Digital Island, lock, stock and barrel. While building a monster network of servers and data
centers to deliver far-flung content may have made sense a few years ago, these days it just
doesn’t make financial sense unless you’re benefitting from the economies of scale a telco
enjoys. Look for more telco/CDN marriages, and soon.
6. Hybrid “Traditional” Content Delivery Networks With P2P Content Delivery At
The Edge Will Begin To Form
Akamai may have the muscular network of nearly 12,000 servers and the greater cash flow, but
it’s United Devices that’s an Internet distributed computing power. We can envision UD
forming Akamai’s new P2P streaming edge.
Sure, UD doesn’t own the computers in their network any more than Blue Falcon does when
100,000 customers spontaneously sign on to listen to Jennifer Lopez on Radio Free Virgin, but
we think larger, more-conservative CDNs like Akamai, RBOCs like Verizon and other “suits”
will see the existing networks as a safer play.
After all, the distributed comp companies at least have some idea of exactly where the
computing edges are and which users, exactly, are signed up to provide the computer.
The fact that the telco/CDN/whomever will have to then acquire or license P2P content delivery
technology as well won’t hold them back from this strategy; the market is growing rapidly
enough that they’ll have their choice of potential vendors/partners from which to choose.
7. Security Issues Raised By P2P Could Remain Critical For At Least The Next 12
Months; Enterprises, However, Need Not Share These Fears
On-demand P2P content delivery can raise some security questions. If a Webcast reserves
chunks of a consumer’s hard disk and passes information along via successive hard disks, only
the most stringent attention to security could lock down the data well enough to please those
with serious security concerns.
On the other hand, enterprises whose PCs behind a firewall may only wish to initiate Webcasts
from within that firewall, and therefore very well may not fear security breaches nearly as much.

For the near term, however, those critics who dislike, fear or compete with P2P streaming do
have some leverage when they complain that there are inherent security concerns in sharing hard
disks for on-demand P2P Webcasting. Even if a P2P content delivery solution doesn’t use this
method, P2P CDN will find themselves defending against this “straw man” in their sales
meetings.
8. P2P CDN Providers Will Need To Own Key Channels Of Their Business
With ChainCast acquiring Internet radio broadcaster StreamAudio and ramping up rapidly in
their delivery channels, we believe the first shoe has fallen. This deal illustrates an important
principle at work -- that P2P content delivery may work better if providers pick up others
downstream in their delivery plans.
We expect to see more acquisitions integrating P2P CDNs with their channels in the near future.
Further acquisitions are likely in the Webcasting arena; we also expect to see P2P content
companies acquire financial fulfillment technology firms (to offer a more complete content sales
option to customers. (Along the way, we also expect to see broadcasters work the deal in the
opposite direction, acquiring P2P CDNs to shore up their distribution mechanisms.)
9. P2P CDN Public Offerings Are On The Way
We think the companies we’ve profiled in this report may be the first wave of tech firms to re-
experience a healthy IPO climate, and we expect them to take advantage of it.
Sometime in the second quarter of 2002, look for a few of the players with positive cash flow
and big-name partners to hit the public markets.
10. P2P Content Delivery Will Find Consumer Acceptance Quietly, But Widely
The P2P content delivery market is maturing more quietly than most, but it has a better chance
of gaining traction than it may appear on first glance, we believe.
Consumers are already familiar with user-to-user computer sharing through networks like
Napster and Gnutella. If explained properly, P2P CDN plug-ins really aren’t hard for
consumers to swallow.
Though we haven’t done primary research in this area -- so readers are advised to treat this as
the hunch it is -- we believe that it won’t be hard to win user acceptance of P2P content
distribution generally, though large client downloads will probably still meet with some
resistance.

With a growing number of intelligent, decently funded companies seeding the market with this
technology, consumer and for that matter enterprise acceptance will firm up with seeming
suddenness within six to twelve months.
Chapter 1 Introduction: The Times, They Are A Changin’
Early on in the Web revolution, content creators discovered that they had a problem. When
they let it be known that a desirable, relatively high-bandwidth piece of content was going to be
broadcast, --- say, a video stream of the desire-inducing Victoria’s Secret models strutting their
stuff in this year’s new lingerie collection --- all hell was unleashed on their servers.
With thousands upon thousands of end users requesting copies of the same content from the
same server group, users experienced slowdowns, denials of service or other annoyances that
didn’t improve their quality of experience one bit.
Before long, hosting firms, ISPs and specialized content distribution services began to offer
specialized methods for delivering both static and streaming Web content to end users more
efficiently.
A couple of major models emerged right away, one focused on accelerating Web server and/or
network performance, the other providing an entire network controlled from end to end through
which to provide a carefully-controlled high quality user experience. Both were intended to
make the unpredictable, delay-prone Web a stable and smoothly-flowing source of content.
Luckily for the future of the Web, these content delivery solutions worked. Users saw much less
latency and better throughput, which kept them hooked and assured that the phenomenal
growth of the commercial Web would continue. Companies like Akamai and Digital Island grew
dramatically as global enterprises signed on to take advantage of their content delivery muscle.
Rapid expansion was typical during the Gold Rush days of the mid- to late-1990s, when it
seemed the dotcom bucks would never go away, and big marketers were spending very big
dollars to establish secure brand names on the Web.
Sure, Akamai might charge $17K, $20K, maybe even $70K a month in unusual cases to carry
a high-quality streaming broadcast. But if you were a product manager with Universal or Sony
or Procter & Gamble, you probably decided that the lower-risk, better known solution was
worth the high price tag.
Still, traditional content delivery players were headed for a major fall. When the tech market
imploded, in fact, the more compromised of the content delivery networks began to teeter
dangerously. Waiting for your long-term “waiting for profitability” strategy to materialize takes a

lot of patience as it is. Having your giant market capitalization vaporized can snatch the ground
right out from under a company with a high burn rate.
Given the damage resulting from the stock market free fall, as well as some underlying issues
inherent in building out an expensive infrastructure, things are changing the content delivery
world.
For that reason, we believe that the P2P streaming companies have at least a six- to 12-month
opening in which to make their case to consumer and enterprise buyers interested in Web
performance. These days, in other words, some folks will take them seriously that might never
have done so if the tech market hadn’t collapsed.
Just how big an opportunity do they have? We project that the Web content delivery services
market will increase from about $404 million in revenue for 2001 to $3.1 billion in 2005,
climbing at a compound annual growth rate of 68%. The CDN services niche will be held back
slightly by competition from infrastructure players in the intelligent caching market, as some key
potential corporate customers choose to build their own Web content delivery infrastructure
over their existing networks.
At the same time, we think it likely that P2P content delivery will shave an increasingly large
share of the pie away from the traditional CDNs from a scant 2% over the next twelve months
to 17% by the year 2005.

After all, there’s the bottom line to drive the business forward even if nothing else. Even if the
P2P streaming price claims are exaggerated, meanwhile, there’s so much “wiggle room” that
P2P streaming is likely to offer very substantial savings even if the real numbers are greatly
reduced.
What’s more, today’s content delivery networks are in a rather vulnerable position, as we see it,
despite some rather formidable strengths on their side as well.

Chapter 2: Traditional Web Content Delivery
Many lessons can be drawn from the tragic World Trade Center and Pentagon bombings on
September 11. During the first hours of the crisis, millions of other Americans tried to reach
CNN.com and other large news sites, but had little or no luck pulling down live content. This
was true despite the support and intervention of traditional content delivery networks.
Coping with peak loads of this intensity is just one of the challenges traditional CDNs face if
they hope to make it into the next phase of Internet growth. While we feel confident that both
these infrastructure-based CDNs and P2P CDNs can and will find opportunity, traditional
CDNs have their work cut out for them.
CDNs vs. caching players
In discussing CDN fortunes, it’s important to separate them from hardware/software players
focused primarily on caching mechanisms as a means of speeding up content delivery (see chart
below).
Table 2.1 Major Web Content Delivery Models
Facilities-Based Content Delivery Services
Examples: Akamai, Digital Island, iBeam, Speedera
Companies like Akamai and Digital Island own some or all of their server infrastructure. Their
strategy is to control broadcasts of content from end to end all the way to the end user’s
desktop. They do this by keeping careful control of the pages served or streams which are
broadcast, making provisions for failover and most often, managing as much of the manner of
content delivery as possible through the use of sophisticated software. Akamai’s FreeFlow
service, for example, offers software capable of serving pages prioritized according to content
type.
To stay in business, these companies must invest large amounts of capital in maintaining their
infrastructures, which may include many thousands of servers, leased backbone connections and
globally dispersed data centers.
Strengths: There will always be a certain price-insensitve segment which, if it considers
Akamai and Digital Island and iBeam to be the top drawer players in the market, won’t switch
to less expensive options casually despite hard economic times. Also, owning a lot of
infrastructure does offer conspicuous control over a user’s ultimate quality of service.

Hardware/Software-Based Caching and Acceleration
Examples: Inktomi, CacheFlow Inc., Persistence, InfoLibria, Cisco Systems, Network
Appliance, Novell Corp.
This group of vendors offers solutions that speed up and improve the quality of Web content
delivery through varied content caching schemes. Depending on the vendor involved, caching
may occur at the server, within the network operating system or at the edge of the network
close to the user. Hardware/software bundles are common, but software-only solutions exist,
as do solutions that rely primarily on a caching “box” placed strategically in an ISP’s rack room.
These vendors aren’t usually selling directly to end users, although very large enterprises may be
large enough in scale to benefit from the efficiencies provided by such products. More typically,
this class of companies comes into a project either to assist an ISP in strengthening its content
delivery or help build a new content delivery service with an existing managed service provider.
Strengths: ISPs and hosting firms on intimate terms with a given piece of hardware are
unlikely to part with it lightly. If a CacheFlow box and operating system are integrated into an
ISP’s network, for example, and the administrator likes how they work, he or she is unlikely to
forego their favored platform the next time around.
According to IDC, the intelligent Web caching market alone is primed to explode during the
next few years, climbing from $514 million in 2000 to $4.5 billion in 2004.
To understand how facilities-based CDNs will compete with the emerging P2P CDNs, it’s
important to take a close look at how the previous generation of CDNs gained their position,
how they currently operate, where they are today and where they seem to be headed.
The buildout race
The content delivery networks that emerged as brand leaders during the first era of expansion --
notably iBeam, Digital Island and Akamai -- went through very similar growth phases, investing
heavily in infrastructure and burning through capital at a frantic rate.
Over a period lasting just about 18 months, ranging roughly from mid 1999 to late 2000,
content delivery networks went through an intense period of infrastructure build-out, scattering
servers and connecting with networks throughout the United States and a large number of
foreign nations.

Looked at from a bulk standpoint, the winner of the content networking sweepstakes was
almost certainly Akamai, which had 8,000 servers in 55 counties in place by the end of 2000
(and has since built that number to 11,600).
During that same period, iBeam co-located its servers in 190 ISPs), including At Home
Corporation, America Online, Inc. and regional Bell operating company BellSouth
Telecommunications Inc. To shunt data from place to place, iBeam put in place a combination
of standard wired connections and satellite broadcasting channels.
Digital Island, meanwhile, could boast 2,200 edge servers and eight data centers touching 260
networks by the end of the year 2000. In its annual report for that year, Digital Island touted its
ability to reach 95% of the world’s business population.
Other firms competing with these players built out overlay networks intended to deliver content
largely over the Net’s existing servers and connections. Companies like EpicRealm, Internap
Network Services Corp. and Speedera introduced technologies and designed to route data
intelligently and fight network congestion.
Emerging markets
During this key period, the content delivery networks were competing intensely for customers.
With the industry exploding, and the solutions sounding rather similar -- even if priced
differently and playing out differently in practice -- the tug-of-war was intense.
Since then, the CDNs have learned some of their key lessons: 1) enterprise clients may actually
offer a brighter future than the originally attractive media and entertainment giants and 2) CDNs
probably can’t achieve sufficient economies of scale to deliver network services unless they
team up with much-larger network service providers and 3) E-business content delivery and
data management may be as important or even more so than handling the throughput of
streaming media or standard editorial and artistic content found on traditional Web sites.
iBeam, for example, had set out to focus on entertainment and media customers. But over time,
it found that enterprise customers were the ones filling its coffers. Enterprises represented 53%
of revenue in 2000, including Accenture (formerly Andersen Consulting), Lotus Development
Corporation, Honeywell International, Inc., Merrill Lynch Fenner & Smith, Inc., Morgan
Stanley & Co., The Goldman Sachs Group, Inc. and Salomon Smith Barney Inc.
For another example, look at Akamai: a substantial minority of its business was generated by
Apple Computer during its first two years of growth, amounting to 22% of revenues during its
first year of business and 12% during its second year.

Financial struggles
By the end of 2000, more than a few of the pure CDNs ended up a thoroughly compromised
position financially, and the year 2001 has not improved their prospects much.
CDNs have generally been unable to raise prices fast enough to keep up with the cost of
providing the elite and highly specialized services they have to offer. Akamai, for example,
found that the cost of service went up 118% during the last six months of the year 2000 alone,
making rising customer volume a further drag on profits rather than a financial benefit. (see
Akamai profile, below). This is true despite the fact that Akamai’s prices are among the top tier
in the industry.
Amidst a host of workout-style tactics, including restructurings, layoffs, and moves to drop
smaller, less profitable customers, it appears clear that being a pure CDN is a very difficult
business in which to achieve a profit, though it may be quite possible to generate a great deal of
revenue.
Over the last several months, a financially tottering iBeam found its salvation in Williams
Communications, a major backbone operator whose network spans 33,000 miles across North
America. Williams took a 49% stake in iBeam. Even with the $40 million infusion that came
with the Williams investment, iBeam’s management isn’t sure the company will stay alive past
the 4th
quarter of this year.
Digital Island, meanwhile, was acquired in full by telco Cable & Wireless. Like iBeam, Digital
Island burned through huge amounts of cash building out its core infrastructure, but the company
has never yet turned a profit. The telco deal bought the company some breathing room, and
margins are improving.
Akamai remains perhaps the largest facilities-based CDN that hasn’t joined with a major
external partner as a means of keeping on its feet.
Pluses and minuses
As these traditional CDNs have lasted out these storms and matured, time has thrown both their
strengths and weaknesses (financial and operational) sharply into relief (see Table 2.2, below)
On the negative side of the balance sheet, forces are now work which might make CDNs
attempts at financial recovery insufficient on their own despite varied cost-cutting measures,
investments and mergers.
Price-cutting pressures are descending from various directions which may undercut the
profitability of CDNs to an unacceptable degree, perhaps even CDNs which fall into the arms
of a brawny networking player like Williams or C&W.

Table 2.2
Infrastructure-based content delivery: pluses and minuses
What’s right with centralized distribution?
_ Ease of use. Generally, nothing needs to be downloaded. Windows Media Player, Real
Player, and Flash are built-in parts of most browsers.
__ It works very well under ordinary peak loads.
__ It can be very profitable for centralized CDN distributors.
What’s wrong with centralized distribution?
_ A centralized distribution network does not degrade gracefully under extremely heavy loads.
_ A structure built to handle peak loads has to be paid for continuously even when the current
load is nonexistent.
_ Heavy existing infrastructure drastically increases charges for centralized CDN customers
who want to deliver content to a large potential user base.
_ This charge structure locks smaller customers (e.g. independent record and film labels) out
of this kind of service.
_ There aren’t that many potential customers who can afford to distribute content this way.
We don’t mean to imply that traditional infrastructure-based CDNs will dry up and blow away
in a matter of months. Given the growth in revenues realized by companies like Akamai,
demand is clearly there -- or at least has been prior to the tech crash, and may very well be
there again as the current U.S. recession heals.
Also worth mentioning, virtually all of the CDNs noted plans to expand internationally, but none
appears to have been highly successful to date. If any of the players now struggling for life were
to find a solid new market in, for example, Internet-friendly Asia, it’s possible the dynamics of
their business could be different -- and reward the costly buildout of their infrastructure which
has bled them dry to date.
We do, however, see it as obvious that the existing CDNs are more vulnerable than they have
ever been to competition from newcomers, especially those that can deliver even a reasonable
facsimile of their product at a lower price. As we’ve noted earlier, we believe that this has
opened up a window of opportunity for the P2P players which might not have existed so much
as six to eight months ago.
As P2P content delivery players go to market, boasting prices dramatically lower than those
possible for CDNs with a massive network infrastructure, we see it as inevitable that significant

downward price pressure will take place. While it’s too early in the game to say whether all of
the cost-savings claims made by the P2P CDNs can be realized -- 90% discounts are a very
tall order -- even the air of credibility is enough to begin to shift theperceptions of buyers, we
believe.
Combine that with an overall feeling among those buying the services that the Internet gold rush
era is over and that “price is no object” philosophies don’t play anymore, and you have a
situation in which consolidation, acquisition or collapse of unprofitable CDNs is inevitable.
However, we don’t think the traditional CDNs’ model is untenable. If the CDNs find a way to
amortize the cost of their infrastructure across a larger network -- whether through partnership,
merger or investment -- they do offer a unique service which seems to be in demand among e-
business players.
Quick uptake of Akamai’s new EdgeSuite among e-businesses points to an unmet data delivery
need -- and security arguments that a publicly-channeled P2P CDN won’t win, at least for now.
Securing corporate data while assuring good throughput will probably be the a key business
model for traditional CDNs that survive.

Chapter 3: P2P Content Distribution: Background and Basics
Over the past eighteen months or so, a new force has hit the Internetworking world, one which
many see as capable of changing the dynamics of content delivery. “Peer to peer” applications
are a new class of solutions which link computing devices together -- usually desktop PCs, but
quite often handheld computers, PDAs and cellphones as well.
In the P2P model’s “pure” form, the peers work together without needing a server to mediate
or control how they interact in any way. In all practicality, most P2P models touch a server at
some point in the architecture, but most still offer edge computing devices a remarkable degree
of freedom from the central computing infrastructure.
P2P companies: A background briefing
Among the list of companies that call themselves peer to peer, several major categories have
emerged.
To provide some perspective, we’ve listed below some examples of the leading players in each
of some of the major peer to peer subsectors (other than the P2P content delivery space,
handled later), and provided a few details on these vendors’ basic configuration.
P2P collaboration
This class of applications allow users to form workgroups and collaborate on the fly, with no
intervention from a mediating server. While servers play a role in this type of app, they usually
aren’t needed as go-betweens to support task and file-sharing between computing devices like
PDAs and handhelds.
•• Groove Networks
Launched by Ray Ozzie, developer of the world-famous Lotus Notes collaborative software,
Groove Networks has built a P2P platform emphasizing secure, spur-of-the-moment peer
group collaboration. Activity between workers takes place in “shared spaces” available only to
invited parties; all data is encrypted.
• Endeavors Technology
Endeavors’ Magi Enterprise product offers file sharing, file searching, instant messaging and chat
between peer devices, which in its current implementation include Win CE handheld devices,
PCs and laptops. When used as P2P infrastructure, Magi makes use of HTTP, WebDAV and
other open standards protocols.
Distributed computing on the public Internet
Internet-based distributed computing schemes tie hundreds or thousands of consumer
computers together across the Internet. Firms in this category resell these “virtual

supercomputers” -- composed of the aggregated spare CPU capacity of hundreds or thousands
of Net-connected user PCs --to companies needing intense computing power.

•• United Devices
The grandfather of Internet-based supercomputing, UD provides an Internet-based computing
utility through its hosted “Global MetaProcessor” service, which draws on the power of UD’s
member network of Internet-connected PCs. In addition, UD offers enterprises its
MetaProcessor Platform for consolidation of internal PC resources.
• Parabon Computation
In addition to providing enterprises with the capability to link their spare computing resources
together, Parabon’s Frontier solution includes an option using the public Internet to offer
supercomputing capabilities to industries with high-intensity computational needs
Distributed computing in the enterprise
Enterprise-based distributed computing systems work like Internet-based schemes, connecting
large numbers of PCs together to produce a massive computer or a distributed application
processed in tandem by multiple computers. However, these platforms only access internally-
owned PCs connected to a given enterprise’s LAN or WAN.
• Avaki Corp.
Avaki’s 2.0 technology aggregates the power not only of desktop computers but also
workstations, servers and clusters together. Their approach is to either provide shared data
access or link these computers into a single computing grid.
• Entropia
Entropia’s solution creates a “virtual supercomputer” pulled together from desktop PCs
connected to an enterprise network. Components include a Network Manager (located on a
server PC), a Job Manager allowing users to submit and monitor jobs and client software for
the desktop PCs.
•• Data Synapse
Data Synapse’s Live Cluster solution can draw on desktop PCs, workstations or servers. The
client machines, are known as “engines,” are coordinated by a server responsible for parcelling
out jobs. A component known as a “driver” communicates between the engine (client machine)
and the server.
• Platform Computing
Platform’s product manages work across servers, desktops, supercomputers and clusters. Its
solution includes: 1) a resource management component which monitors distributed resources,
responding automatically to increase app and services availability and 2) a performance
management component offering enterprises insight into workload and system performance.

P2P content management
•• NextPage
NextPage’s products allow employees, partners and customers to access distributed content,
create and store business processes, and collaborate with peers. Its product line includes NXT
3, a platform tying distributed content servers together; NextPage Matrix, which enables
business processes and where needed, captures them; and NextPage Solo, which provides off-
line access to the NXT 3 content network.
•• WorldStreet
WorldStreet offers the World Street Net, an infrastructure enabling users to manage and
automate content sharing with others - including colleagues, partners, clients, and suppliers -
based on the specific nature of their relationships with these individuals.
•• Xdegrees
Xdegrees is a networking company offering its own method for tracking content and other
resources on the Internet and corporate networks. Its eXtensible Resource Name System,
which gives resources addresses similar to IP numbers, allows companies to access files,
services and machines securely from any connected point on the Internet or network.
P2P content delivery
As P2P development has gathered steam, P2P methods for delivering Web content have
emerged. They vary distinctly from “content management” vendors such as Next Page,
described above, in that they go outside the corporate network to get their job done, physically
piggybacking onto infrastructure enterprises don’t own.
Along the way, each of these companies intends to offer corporate content creators and
Webcasters a substantial savings over traditional content delivery methods. These players say
they can offer lower prices because their approaches consume far less bandwidth, server
resources and support infrastructure than technologies employed by their traditional CDN
peers.
How do the P2P CDNs propose to offer such a big savings? While their technologies differ in
the specifics of how they get the job done, they’re more or less all talking about using user PCs
to pick up and broadcast streams or down load to their neighbors. Rather than servers doing
all of the content pumping, servers distributed once and PCs pump to PCs. Below, we’ve
provided an overview comparing the standard process of server-intensive content distribution
with the emerging P2P content delivery network model.

Table 3.1
Standard Web Content Delivery vs. P2P Content Delivery
Standard Web Content Delivery
1. A user sits at their computer and makes a Web content request.
2. The request comes from across the Internet. If the user requesting the information is not
close to the server geographically, the request will “hop” from router to router until it
reaches the server in question.
3. The hops it makes are following the path the router has determined to be most efficient,
based on its current “understanding” of Internet topology.
4. Web content is pushed out by a server, a computer whose job it is to send along a
document or other identifiable piece of data to a user upon request.
5. The user receives the HTML-coded content and can view it with his/her browser.
P2P Web Content Delivery
1. A user sits at their computer and makes a Web content request. The request is handled by
special P2P content management software already downloaded to their machine.
2. If the user is very close to the server, the user’s request may end up going to directly to the
original server across the Internet.
3. If the user is further away from the server (more likely) User #1 will be pointed by directly
to the PC of another user who has downloaded the content requested. The “host” PC is
chosen by the topology-aware software User #1 has already downloaded, and typically is
intended to be as close as possible geographically.
4. The P2P system may try to match User #1’s bandwidth with User #2’s bandwidth.
5. User #1 downloads their content item. If User #2 happens to sign off during the download
-- as she/he may or may not be aware of the download -- P2P Web content systems
typically switch User #1 over to another host machine instantly.
Of course, in practice, the various companies attempting to bring P2P content delivery down to
earth have come down with a wide variety of interpretations of this scheme outlined above.
Among the key distinctions come in how the companies manage digital content (existing
solutions from, say, Microsoft vs. their own), how they deal with any failure in the content
network and how they actually deliver the content delivery.

Over the coming pages, we’re going to acquaint you with the range of P2P content players
which have emerged to date and explain how they differ, both in their attempts to monetize the
P2P delivery model and in their basic technical underpinnings (see chapter 4 for more details).
Categories: P2P content delivery
Generally speaking, there are two categories of P2P content delivery provider, though newer
models with unique features have begun to emerge over the last quarter. It’s important to note
that some of the P2P CDN competitors offer both streaming and download-oriented services
•• Streaming delivery
Companies offering streaming P2P content delivery have developed a variety of technologies
designed to superimpose a virtual content delivery network over a spontaneously-formed of end
user PCs equipped with client software. These companies, generally, intend to make money by
helping enterprises, entertainment companies or network service providers reduce their content
delivery costs.
Competition:
_ Traditional audio and videoconferencing providers
_ Infrastructure-based streaming services providers
_ Infrastructure-based content delivery networks such as Akamai and Digital Island
_ Possibly, traditional radio and television, to the extent that traditional streaming services
offer any competition to these existing media
•• Download delivery
Download-oriented P2P content delivery providers are emphasizing “time-shifted” delivery, in
which the user downloads a larger file for later use (in the case of software) or viewing (in the
case of video or possibly audio files). Their bet is that users are willing to wait for downloads
and pay a fee to see high-quality entertainment or use software.
Competition:
_ DVD, CD- or videocasette-based entertainment products (movies or multimedia games)
_ DVD or CD-delivered software products
_ Entertainment streamed by traditional means (e.g. Digital Island, Akamai, Speedera)
_ Entertainment delivered via traditional radio, television and cinema venues

__ Other/hybrid
At least one company, Uprizer, is looking to get content out to mobile wireless devices using a
secured P2P content delivery approach. (See profile, below, in our Appendix.) Uprizer’s
approach is complex, but among its notable features is a proposed service charging a monthly
fee to provide downloadable clips of entertainment-oriented video -- for example, a key play in
a football game -- to a user’s mobile device. Uprizer does not propose to “time shift “ in a
mobile environment, however; Uprizer anticipates delivering content instantaneously on demand.
Competition:
_ Text-based entertainment or news-oriented services delivered via wireless carriers
_ Streaming services delivered by traditional players
_ Streaming services delivered by P2P players
Technology comparison
When comparing these companies, it’s helpful to separate out their approaches to key portions
of their technology. In this paper, we do not propose to undertake a highly technical analysis
these products, but we do hope to familiarize readers with basic functional differences between
them.
Below, we offer a review of key features among a few P2P streaming content delivery players:
AllCast
__ Technology: The AllCast Broadcaster, a Windows-based application for broadcasting live
with P2P technology. The Broadcaster distributes bandwidth usage among the P2P
participants in the broadcast network.
_ Client side: Users connect using the AllCast Plug-In, which integrates with the Windows
Media Player. Includes a “chat” function.
__ Network side: The Live Broadcasting Server manages the connection structure among the
users connected to the network, performing live data compression and then broadcasting
the data. It also offers supplementary data transmission in order to display advertising
information.
__ Management tools: The Plug-In provides statistical information about current P2P
conditions via a taskbar icon. The Broadcaster software also provides content delivery
information.

ChainCast Networks
_ Technology: ChainCast’s core technology is the Network Topology Engine, which keeps
track of online listeners in real time as they connect and disconnect. It can handle up to
500,000 end users connected to a given content source. The NTE’s job is to determine the
best content source for a given client computer and ultimately, to optimize performance of
the entire virtual network it has created.
__ Client side: To participate in the ChainCast Virtual Network, users download the VMR
(Virtual Multicast Router) Player.
__ Network side: The VMR Server receives audio content from a Webcaster, encodes it,
and transmits the stream over the ChainCast Virtual Network.
__ Management tools: The ChainCast Manager offers control of the virtual network through
a central operations point. It can offer origin server signal detection, system fault monitoring
and event management, virtual network topology recordings, performance monitoring, data
collection and management and security.
VTrails
_ Technology: Uses “Full Duplex Packet Cascading” technology to deliver streams across
networks of user PCs. When receiving IP requests, the vTCaster maps the network context
of the end users based on their network location and connection type. vTCaster then builds
the most efficient cascade delivery path, prioritizing the order in which the end users are
served and connecting users coming from the same network, ISP, or organization. vTCaster
keeps track of end users joining and leaving the stream, and dynamically updates the
cascade delivery path.
_ Client side: To participate in the network, users download vTPass client software, which
integrates with standard media players.
_ Network side: The vTCaster box sits at the broadcast source and processes IP events.
When needed, the vTEdge server will be implemented to further remove load from the
network and content closer to the end user.
_ Management tools: vTrails offers VTReport, an application tracking unique stream
requests, regional information, stream duration, usage count and usage habits, and
vTControl, a broadcast management app offering control over the vTrails System.
Conclusions

P2P content delivery is a unique service which differs substantially from other emerging P2P
services and products.
For the short term, streaming services and download services are likely to take different paths:
_ Streaming P2P players will be plugging themselves into the Internet side of the broadcast
world or competing/cooperating with corporate audio/videoconferencing service providers
_ Large-download oriented P2P players will form partnerships with the big boys creating
large-file style content (film-makers, software developers, music video producers) and if
effective at creating a revenue stream, get acquired by those players.
At the moment, a few companies are trying to offer themselves as experts at both sides of the
equation, but we don’t think this positioning will last. Ultimately, we think these markets will
split further apart, at least if they keep the character we’ve outlined, because as outlined
companies in each sector are really going for different results.

Chapter 4: Markets And Models
The P2P CDN market (and, we believe, the infrastructure-based CDN market in all likelihood)
is well on its way to splitting into a handful of market segments, each of which will call for its
own approach and entail its own risks. Below, we’ve provided some details on the
opportunities and challenges we see for each niche:
Entertainment/media
The P2P or non-P2P player that delivers quality streaming, on-demand streaming or even file-
downloaded entertainment (think movies) or broadcast content cheaply could potentially make
a fortune. On the other hand, the market is still very volatile, and it’s still not clear quite how
consumers want to buy.
__ Business model
At least in some sectors, entertainment buyers are used to paying a modest fee for content they
can only access briefly (e.g. video rental, movie theater shows, game rental) or items they can
download (shareware and some types of software). Selling technology to the creators of such
content can result in big bucks for vendors if the fees sharing/licensing deals are structured
correctly. The software market, meanwhile, has long been waiting for an efficient way to deliver
its product via rights-managed downloads that don’t turn off buyers.
__ Challenges
Many other forms of entertainment come in packages which don’t require a download and
persist in a form (CDs and videocassettes) with which users are already familiar. Getting them
to pay for entertainment delivered to a computer -- which may lack the multimedia capabilities
to do such content justice -- is far from a lock. Software buyers, too, can always resort to their
CD player.
Enterprise content delivery
As enterprises become accustomed to using their PCs in distributed computing networks, we
believe that they will be quite receptive to considering P2P streaming and management of key
downloads as well, at least within their own networks. As large as the entertainment/media
niche is, this market will ultimately outstrip it.
__ Business model
There are at least two existing niches within the corporate market for P2P content delivery.
Companies may buy streaming P2P services as an alternative to traditional streaming or even
network-based video- or audioconferencing. Enterprises may also use the download vendors’
capability to deliver e-training to employees across their LAN or WAN.
__ Challenges

Change happens and new approaches get adopted, but often too slowly for technology
vendors. While we think it’s inevitable that the enterprise P2P market blossoms, we think it
could take as much as 18 months before large enterprise begin to take a serious look at this
approach. In the meantime, some lesser-funded P2P CDN players who could have won in this
space may fold instead.
Network service providers
P2P technology developers are beginning to license their technology to network services
providers of various kinds ( ISPs, ASPs and other xSPs), with the expectation that they will in
turn offer its capabilities as a managed service. It’s still up in the air how to charge for these
licenses, but it’s clear that turning the tech over to networking companies can be a beneficial
deal for both sides. The P2P CDN developer gets the cash, and the NSP gets a means to
deliver a valuable service cheaply, produce happy customers and presumably still make a profit.
__ Business model
In existing models, network service providers license P2P content delivery software on a per-
user or per-MB-delivered basis, filling the coffers of vendors while simultaneously growing the
overall market acceptance of P2P CDNs generally. In other instances, the xSP may buy the
technology outright, with similar market-building results.
__ Challenges
Few. This is perhaps the most promising market for P2P CDNs if they wish to maintain their
current configurations as technology vendors. The only real question is how to best build on the
NSP relationship to further develop the software; that may call for a closer partnership than
many of the “mile-wide, inch-deep” relationships appearing in press releases, and those take
time to develop.
Telcos
Carriers eager to sell their broadband services might find cheap streaming or download delivery
to be a potent incentive to sign on -- that is, at least, what some of the P2P CDNs are banking
on as an initial strategy. We see their logic as compelling, and think telcos like AT&T -- with its
cable assets along with traditional telco infrastructure -- as very logical players in the streaming
market, especially if it has a new technology in hand which can overleap entrenched players like
the infrastructure-based CDNs.
__ Business model
Models here resemble models for network service providers, described above. But given that
carriers in this instances are after consumers and small businesses, strategies for licensing
software are likely to be different. Business models will probably revolve around offering the
software as a packaged product for hosting in the carriers’ own network operations centers.

__ Challenges
Even a deep-pockets carrier may balk at being constrained by per-MB charges if they’re
hoping to encourage high use of broadband services in homes and small businesses. P2P
content delivery vendors will probably need to be creative in structuring fees if they go the
licensing route. As for software sales, we think the main issue will be overcoming the
conservative carrier mindset, which may distrust the P2P virtual network even if its engineers
support a vendor’s claims.
Enterprise data delivery
As the success of Akamai’s EdgeSuite service has demonstrated, companies are willing to pay
heavily to have bits from their data-intensive Web sites delivered reliably. It’s not surprising --
success in e-commerce, of course, can depend heavily upon the timely delivery of such data.
We realize that such customers may be loath to trust the security of such deliveries, but it’s
beginning looks like secure transmission of P2P data is possible. One company not profiled in
depth in this report, QuMatrix, apparently developed such secure P2P technology that an
international bank is now using its technology to transmit financial data.
__ Business model
If a P2P CDN can offer airtight security, businesses are likely to pay a premium to distribute
critical data, particularly across trusted environments such as intranet or locked-down supply
chain architectures.
__ Challenges
Security is as much about politics as it is about science. As long as the content delivery is within
a corporate network, it will be easier to make and win an argument that security is adequate.
Once a vendor tries to promote P2P-relayed content delivery of secure data, they’ll face the
same concerns that Internet-based distributed computing companies like have -- which is to
say, worries which will require a lot of reassurance and lots of technical backup to calm.
Conclusion
We think the market has a ways to go in sorting out pricing issues -- there’s a great deal of
wavering going on among vendors -- but that will sort itself out within the next six months, we
predict.
Another area which still shows fuzzy outlines is the conflict between serving as a managed
service provider and serving as a software/technology vendor. Vendors interviewed for this
research seem convinced that they can work effectively cultivating both managed service
relationships with clients and selling software or other technology to end users. But based on
our experience with other network-based services, we’re skeptical that the two strategies can
coexist comfortably.

We think that the P2P CDNs will have to take a stand as to their mission. They’ll have to
decide whether they’re going to sell raw technology (in which case they’re probably better off
going nuts building software-industry partnerships and plugging themselves into corporate
solutions) or service (in which case they need to really focus on building out data centers or
partnerships with bandwidth providers and so on.)
In covering these various niches, we haven’t gotten into some emerging wireless content
schemes which may soon be important in the P2P content delivery world. For more detail on
one such scheme, see the Uprizer profile in the profiles section of this report, below. The
sectors listed in this chapter, however, represent the P2P CDNs’ main areas for growth in the
next 24 months, we believe.

Chapter 5: Forecasts
Savings will trump all
We’d argue that for a large segment of technology buyers -- including enterprises, broadcasters
and entertainment companies -- price is more important than it was before. For that reason,
P2P content delivery will look awfully attractive. Any technology boasting savings ranging from
50% to 90% over traditional Web content delivery has got to have the ear of a great many
content creators.
At this point in the discussion, we’re not engaging in arguments as to whether the results
produced by P2P streaming or on-demand offerings are of an equal quality to those produced
by traditional CDNs. We’re only noting that if something’s cheap enough, there’s usually a fair
number of folks who’ll decide that it’s good enough.
P2P/traditional CDN hybrids will emerge
Before the traditional infrastructure-based CDNs give up, we think they’re going to make their
truce and join the force -- P2P content delivery players -- they know they can’t beat. For
CDN infrastructure players too cash-strapped to keep up the endless server buildout (and that
includes pretty much everybody in the first generation) P2P content delivery may be a way out.
The user-PC-to-user-PC architecture underlying P2P content delivery (described in detail in the
next chapter) is certainly frightening to companies used to centralized control of content delivery.
The idea of ping-ponging content across a thousand or a million user computers, each of which
is both a client and server, may very well scare the heck out of the leadership of companies like
Digital Island, iBeam and Akamai.
Bear in mind that we’re not scoffing at conservatism; we started out skeptical too. But we now
believe that the empirical evidence will convince skeptics that the leading P2P players do have
sufficient control of failover, fault tolerance, and general network scalability to produce quality
results in real time if they wish. Ultimately, we believe, the traditional CDNs will make the same
decision, for reasons of cold, hard survival if nothing else.
When they do, traditional CDNs will begin to employ P2P content delivery networking
techniques rather at the edge of their networks rather than bulking up on servers. They may turn
to distributed computing companies such as United Devices first to test the application in a real-
world distributed environment with scores of connected end-users, but over time, they will
employ the technology, we contend.

Look for quiet beta tests over the next twelve months and eventually, a fairly noise about-face
by traditional CDNs as they come to acknowledge the benefits of P2P content delivery at the
network’s edge. Also, look for consumer infrastructure companies with less ego invested in
proving the P2P CDNs wrong -- think AT&T, for example -- to step up to the P2P CDN
business.
IPOs are on the way
We think the companies we’ve profiled in this report may be the first wave of tech firms to re-
experience a healthy IPO climate, and we expect them to take advantage of it. Sometime in the
second quarter of 2002, look for a few of the players with positive cash flow and big-name
partners to hit the public markets.
Why? First of all, here’s our feeling about public markets and technology, which you may want
to bear in mind through the rest of this section. We believe that the best time for a technology
company to have an initial public offering is not when it’s at its most mature stage, but when it
has established itself at a certain post-startup, pre-maturity level. Perhaps we could compare
this to getting second-round VC funding. We know others may think of IPOs as appropriate
for mezzanine-style cash, but we think it’s more appropriate to sink dollars into a company
when dollars can have a slingshot effect, boosting companies rapidly up into the stratosphere.
That being said, we believe that the markets and a few of the better positioned P2P CDN
companies fit the profile of tech companies which can benefit from an IPO and represent a
reasonable investment.
The leaders in this space are offering a promising service which:
_ Offers substantial savings over alternatives
_ Attracts companies with large war chests
_ Already boasts developed and rapidly refining technologies
_ Is beginning to attract big-company interest
_ Sits on top of a growing overall market (see our projections in Chapter 1).
In sum, all of the signs are there for a public markets surge. If a P2P CDN player can also
demonstrate upward mobility of income and a short time frame to profitability (i.e. about one
year), we think they’ll be set to go -- and will.
P2P CDNs will acquire infrastructure, channels
P2P CDNs are moving into a new, better defined stage right nowÖor at least they should be, as
indecisiveness will be decisively punished in the marketplace next year.

As mentioned above, they have at least two apparently viable business models, to go either into
the managed services business or the technology sourcing business. A third, which came into
focus when ChainCast acquired Internet radio broadcaster StreamAudio.com, is to build out
into becoming more of a full-service provider offering not only technology and channels, but also
distribution and possibly added services.
If buying an Internet radio broadcaster is a good way of increasing your reach, wouldn’t buying
a financial fulfillment vendor be a great way of increasing your attractiveness to content vendors?
Rather than building a server farms or partnering, even with intelligently-managed hosting
companies, why not buy hosting companies with smart technologies?
What about merging with one of the companies (e.g. Speedera) that specialize in optimizing
networks? The combination of off-the-edge services and a smart network could be a winner.
In short, expect to see these and other border-expanding transactions happen for companies,
like ChainCast, with the financial backing to make them happen.
Security will be a 12-month issue
We predict that security will come up as an issue for P2P CDNs, but will only pose a real
barrier to growth in instances where data is particularly sensitive or users are particularly
paranoid.
Under the covers, many of the P2P content relay schemes tread lightly on a PC, and some
don’t even access the hard drive at all. That takes away some key security mumblings
regarding the hackability of user hard drives. The distribution of files into chunks across a PC
network, meanwhile, only makes it less likely that an intruder would be able to steal copyrighted
content or data.
Still, for the next twelve months or so, we think that security will be the issue most frequently
raised by potential buyers with objections, at least for consumer applications.
There’s something about the notion of streaming across PCs across the great wide world of the
Internet that unnerves content creators and network managers --perhaps rightly in some cases!
-- and it may take more than logic to calm them down.
Luckily for the health of P2P CDN players, corporate entities with secure LANs and WANs
have less to fear from a streaming P2P solution at the outset. Streaming from machine to
machine within their network may seem like a reasonable solution immediately.

All told, we predict that PC security will remain an often-raised issue -- even if it’s a “straw
man” that doesn’t apply to a given technology -- for about the next year, but will die away as a
persistent buyer concern by 2003.

Profiles of traditional content delivery players
Akamai Technologies Inc.
500 Technology Square
Cambridge, MA 02139
Phone: (877) 425-2624
www.akamai.com
History
Founded in August 1998 -- with $8.4 million in from VC firms Battery Ventures and Polaris
Venture Partners -- Akamai is probably the oldest and largest of the pure content delivery
networks.
Akamai introduced its services commercially in April 1999. Since its inception, Akamai has built
out a network comprising more than 11,600 servers touching 820 networks in 62 countries.
No other content delivery player can claim that kind of muscle. Akamai’s strategy, which has
been to bring its servers to more or less every neighborhood its customers wish to reach,
appears to have been largely realized.
While Akamai has attracted a blue-chip client-list cutting across many key vertical industries,
Apple Computer has played a particularly important part in its growth, representing 22% of
total revenues for the year ended December 31, 1999 and 12% of total revenues for the year
ended December 31, 2000. Yahoo was another important early customer, representing 13%
of total revenues in 1999.
Other Akamai customers have cut across the media, news and corporate worlds, including
Barnes&Noble.com, E!Online, MSNBC, Office.com, Progressive Insurance, JCPenney,
Lycos, Martha Stewart Living Omnimedia, Reuters, CBS, C-SPAN, Xerox and The Motley
Fool.
Over the heady years of the Internet boom, Akamai was able to bring a wide range of content
distribution or enhancement technologies in house through strategic acquisitions.
Acquisitions included InterVU, a publicly traded company offering automated streaming media
services for live and on-demand video and audio over the Net; Network24 Communications
Inc., an Internet broadcasting software and services firm; CallTheShots, a company focused on
Web site personalization and content aggregation; and Virage Inc., a firm developing digital
video management software for the Net.
These acquisitions were fed by proceeds from the company’s highly successful stock offering,
which took place on October 29, 1999. Climbing almost immediately to a high of nearly $350

a share, Akamai’s huge market capitalization supported the company’s ongoing need for
infrastructure expansion.
Services
Akamai seems to have been able to integrate its technology acquistions relatively quickly and
well, into services which now form parts of its core product set. Akamai’s current core services
include the following:
Content Delivery Services
EdgeSuite: Positioned to appeal to enterprise customers, this is Akamai’s package of edge
services through which companies can control how data and applications are assembled
presented and delivered to end users.
Streaming Solutions
FreeFlow Streaming: Akamai’s streaming content delivery service.
Akamai Conference: Interactive Web broadcasting solution targeted to business functions such
as training seminars, corporate announcements, production introductions, and investor updates.
Akamai Forum: Going head to head with traditional corporate audio/videoconferencing
providers, Akamai is providing an automated event system which allows its customers to
provide an audio or video Web conference which can include slides and on-demand delivery.
Akamai has had particular success with the launch of EdgeSuite, a recent addition to its service
lineup, which has attracted 75 new customer contracts, including BestBuy.com, Monster.com,
Novartis Pharmaceuticals Corp., Six Flags.com and Victoria’s Secret.
Financial Standing
Akamai’s financial situation is compromised. The company’s expenses continue to climb, as it
invests in forward-looking R&D and builds out its infrastructure further. More critically, the
cost of delivering service has gone up considerably as well.
According to the company’s SEC filing for the 2nd
quarter, revenue climbed dramatically, to
$83.3 million from $25.3 million during the same period in the year 2000.
Still, the cost of service climbed 118% over the same six months, more than doubling from
$29.8 million compared with $13.7 million for the same period in 2000. According to the filing,
cost of service went up due to higher bandwidth costs as a result of expansion of the Akamai

network and “an increased amount of content delivered over our network,” according to the
filing.
This stood out for us. If increasing the amount of content a content delivery company sends
across its network does not result in increased profits, something is wrong with this picture. This
is particular striking given that Akamai is known as being one of the pricier companies in its
domain; if anyone should be charging enough to be making a profit, they should. At the very
least, the company is in an awkward moment in its growth.
Akamai’s stock, meanwhile, like that of all Internet companies, has taken an incredible beating
since its peak days in 1999. As of this writing in late August, it stood at $4.44 per share, down
approximately $340 per share from its peak in the year 2000.
Current Situation
If the initial success of EdgeSuite leads to longer-term commitments from key corporate
customers, Akamai is in as good a position as it could be given its overhead. It still remains to
be seen, however, whether their current business model is fixable, given the company’s reliance
on playing the infrastructure game.
If international telcos with much better economies of scale find it challenging to build out a
hosting network, we doubt Akamai’s smaller network can pay for itself simply through the sale
of network services.
Another, sadder factor may also affect Akamai’s future. Just recently, the company tragically
lost its chief technology officer, Danny Lewin, to the terrorist attacks on September 11 of this
year. It’s not clear yet whether the company’s progress will be impeded by the loss of Lewin, a
well-credentialed and respected technologist.
Net/Net
Akamai has built an incredible infrastructure, appears to have a fine brain trust and has acquired
lots of finely crafted intellectual property. On the other hand, the company’s burn rate seems to
be increasing and its chances of profitability dropping as its expenses increase and competition
from peer to peer content delivery players -- and as-yet unnamed but sure-to-appear additional
competitors -- smell blood in the water.
As we see it, P2P content delivery players will begin to pick off some of Akamai’s less visible
customers -- those more concerned about saving a buck than buying from the corporate
standby which Akamai has become -- and that depending on how Akamai reacts, the process
could accelerate over the next 12 months.

In time, P2P content delivery may even pick up some primo customers Akamai will hate to lose.
In that case, It won’t be the little P2P companies themselves doing the cherry-picking, either; it
will be the larger, more muscular telco and ISP partners recruited by the P2P companies which
will make the more juicy propositions.
We say, hey Akamai, you really can’t ignore P2P content delivery. It’s going to happen, and
you can build on it faster than anybody. Get on board.

Digital Island
45 Fremont Street, 12th Fl.
San Francisco, CA 94105
Phone: (415) 738-4100
www.digitalisland.com
History
Digital Island began life in 1996 as an independent company. Their hook was promising to
speed up content delivery on the Net by shifting traffic off overcrowded main arteries onto its
private “overnet.”
Attracting strong investment interest early on, Digital Island raised $11.5 million to support its
vision within its first year of existence. It followed up the fundraising success with $10.5 million
in March of 1998 and yet another $10.5 million round in September 1998.
In December 1999, it merged with content delivery networking pioneer Sandpiper Networks.
Working together, the two went after major content players on the Net, early on signing a
content delivery agreement with America Online. The company also went after networking
business, for example cutting a deal with cable networking player Road Runner under which it
would speed up content delivery in Road Runner’s key U.S. markets.
The merged companies began to acquire. Early in 2000, Digital Island acquired live and on-
demand streaming media provider Live Online, which it picked up for $65.9 million and $5.2
million in cash. Later that year it bought distributed hosting provider SoftAware Networks for
$390.5 million in stock, $20 million in cash and $3 million in stock options.
Still, financial concerns continued to trouble the company. Such expense worries very well may
have been the reason why Digital Island agreed in May of this year to be acquired by telco
Cable & Wireless. (The deal closed in August.)
Services
2WayWeb
-2Transact
This is Digital Island’s e-business hosting product. Services include two content-related
offerings:
1) Secure content delivery of shared Web content

2) Digital Rights Management options providing digital media content creators with the ability
to sell their wares on a pay-per-view, pay-per-download or subscription basis.
(Subscription option is due to be available in 2002.) Rights management based on Windows
Media Rights Manager 7.0; Real format should be supported next year.-
- 2Manage
This is a suite of services streamlining the process of delivering e-business content, including
Web management and reporting features. Facets include the following:
1. Geographic intelligence services
2. Traffic and performance reporting
3. Web site, application and server monitoring
4. Intelligent traffic management services
5. Online service center
6. Digital content preparation
7. Digital content management
8. Digital rights management
9. E-learning services
10. Streaming media, audio/video production and encoding
11. Secure corporate communications
- 2Deliver
This group of services is Digital Island’s content delivery service, supporting both static and
dynamic content. Services include the following:
1. Streaming audio and video support
2. Distributed hosting
3. Premium network service
4. Content peering, in which Digital Island links up with content delivery infrastructure of other
content service providers to increase performance and support interoperability.
Footprint
Emerging from Digital Island’s merger with Sandpiper Networks, Footprint is the technology
enabling Digital Island’s 2Way services.
Financial Standing
Perhaps due to the intervention of C&W, things actually appear to be moving in the right
direction for Digital Island. Improving gross margin and steadying costs point to a reasonable
hope of profit in 2002, at least when compared to competitors such as iBeam.

Working with C&W seems to be improving the level at which it can utilize this network,
however. Unusual for its sector, Digital Island’s gross margins improved 64% for the nine
months ended June 30, 2001 as compared with the nine months ended June 30, 2000. Still, it is
still sitting on top of a very substantial fixed infrastructure nonetheless, with roughly 2600 hungry
server mouths to feed.
Working to keep the margins healthy, Digital Island instituted staff cuts in May of this year,
saving $6.7 million for the three months ended June 30, 2001 compared with the three months
ending March 31, 2001.
Current Situation
Digital Island seems to be transforming itself from a primarily infrastructure-driven player into a
more flexible company offering content- as well as transaction-driven edge services of various
kinds.
As the growing Digital Island has gained resources, it has attracted a long list high-profile
customers including AOL, Schwab & Co., CNBC.com, Glaxo Wellcome, Microsoft, Sony
Entertainment and Universal Music Group.
The 2Way Web Services launch, a repositioning of its product line emphasizing its high-end
content services over raw infrastructure power, attracted customers such as Walt Disney
Internet Group, Pearson and Cisco Networking Academy. Digital Island was also chosen by
Microsoft to stream the launch of its Windows XP operating system in late October.
Still, ultimately, it’s making a profit that matters, and Digital Island hasn’t gotten there yet,
though it hopes to do so in early 2002. What’s more the cost of revenues has continued to
increase all the way through 2001, a scary sign for a company that hopes to make money
sooner as opposed to later.
Net/Net
Among the infrastructure-driven CDNs, Digital Island seems the most focused, and its financial
trends are somewhat brighter than the others profiled in this report. With the help of C&W,
Digital Island is certainly in the game for the long term.
What remains to be seen, we believe, is whether a meaningful number of Digital Island’s key
customers will be stolen away by less-expensive P2P streaming options as they prove
themselves in the marketplace.

If P2P content delivery can delivery strong results there’s fewer and fewer reasons for
companies to pay premium prices for high-end infrastructure-based services. As we’ve noted
elsewhere, it could be that companies like this one will actually need to adopt streaming and
download-oriented P2P content delivery at some point in the next 12 months to stay
competitive in price-sensitive segments.

iBeam Broadcasting Corp.
645 Almanor Avenue, Suite 100
Sunnyvale, CA 94085
Phone: (408) 523-1600
www.ibeam.com
History
IBeam began its life in October 1999 as a streaming media company focused on delivering
entertainment content.
Like its competitors, iBeam was to find that sales volume was much easier to generate than
profit. From March 1998 through June 30, 2001, iBeam had accumulated a $609.6 million
deficit.
The company was headed for the end. But in a nick of time, a financially tottering iBeam found
its salvation in Williams Communications. Williams, a major backbone operator whose network
spans 33,000 miles across North America, wanted to beef up its streaming capabilities. And
iBeam, having run ashore during a tech recession, badly needed cash.
Williams took a 49% stake in iBeam, putting up $20 million in cash and $10 million in services
to the streaming media firm. Other parties pledged $10 million in cash as well, bringing the total
to $40 million in investment.
Services
Streaming Communications Services Group
This group with iBeam enterprises to help plan and optimize infrastructure for streaming media
within corporations.
Activecast
This is iBeam’s interactive Webcast product, offering a combined package of video, audio,
presentation slides, dynamic Web links, live chat, audience polling, advertising and e-commerce.
TalkPoint
A Webcasting tool which allows users to offer on-demand or live streaming to a dispersed
audience. Audiences view presentations on a browser and listen to audio over either a
telephone or a PC.

Advertising Agent
This is iBeam’s streaming media advertising service. Users can insert live full-motion video or
audio advertisements into streams.
Digital Rights Agent
This service allows users to keep control of copyrighted digital material, either downloadable or
streaming. Makes useof Microsoft Windows Media Rights Manager to encrypt content for
management until the content is sold under terms the creator approves.
Geo Agent
A service allowing content providers to tailor ad delivery to an end user’s location.
Pay Per View Agent
Handles registration, reporting and payment processing for pay-per-view content.
Syndication Media Agent
A hosted application allowing companies to manage distribution of digital media to affiliated
Web sites.
Financial Standing
In recent times, iBeam has fought hard to regroup, discontinuing service to almost a quarter of
its less-profitable and higher-risk customers while cutting back on its employee roster.
Dropping about 100 staffers over the 2nd
quarter of 2000, the move cut its engineering staff by
more than two thirds.
iBeam has managed to lower its cost of revenue slightly from first to second quarters of this
year, in part through staff cuts but also due to a drop in revenue between those two quarters.
The drop in revenue, possibly sparked by fears as to iBeam’s future, doesn’t bode terribly well
in a market where Akamai continues to ramp up revenue.
Even with the $40 million cash infusion that came with the Williams investment, iBeam’s
management isn’t sure the company will stay alive past the 4th
quarter of this year. The
company still isn’t in a very good position to slow its burn rate, and $40 million doesn’t buy a
cash-hungry networking business much time.
Current Situation
iBeam remains on shaky ground financially -- as of mid-August the company was fighting to
keep its NASDAQ listing -- but the company has at least succeeded in shifting its emphasis to
the more-predictable consumer market.

For the year 2000, enterprises represented 53% of revenue. Companies included Accenture
(formerly Andersen Consulting), Lotus Development Corporation, Honeywell International,
Inc., Merrill Lynch Fenner & Smith, Inc., Morgan Stanley & Co., The Goldman Sachs Group,
Inc. and Salomon Smith Barney Inc. By mid-2001, enterprise customers represented 65% of
revenues, and entertainment customers only 35%.
Net/Net
iBeam seems to have evolved into one of the better-operated infrastructure-based CDNs, but
as noted, even the company’s own leadership isn’t sure if it can keep paying for its cash-burning
infrastructure.
iBeam’s executives are now focused on offering business services, and this orientation can do
the job of keeping them afloat, we believe. The question, which remains open at the moment, is
whether business will adopt this alternate model of networked audio and video delivery quickly
enough to support a company iBeam’s size.
Frankly, at the moment we doubt that the cash will move in quickly enough, but we don’t doubt
that the market is there to support a company with iBeam’s profile if things go right. If iBeam
can achieve very substantial enterprise sales penetration -- and it has been doing well with
enterprises of late -- it may just be one of the infrastructure players that survives into the next
generation.
Especially if it finds a way to embrace rather than compete with the P2P CDNs, there’s clearly
a place for iBeam, a CDN that understands how to deliver content to businesses.

Profiles of P2P content delivery players
Acceloop Inc.
1701 W. Northwest Highway Suite 240
Grapevine, Texas 76051
Phone : (866) 222-8140
www.acceloop.com
History
Founded in 2000, Acceloop has focused tightly on content delivery software. Organizationally,
it has a familiar profile for companies with an Israeli arm. The company has declared its
Grapevine, TX location but maintains an R&D center in Nesher, Israel.
Acceloop made the beta of its Web server accelerator available in early 2001 and launched a
commercial version for IIS Web servers later in the year.
Services
Primarily a software infrastructure provider, Acceloop enters the P2P content distribution arena
obliquely. Acceloop’s Web Server Accelerator, which starts at $2990 for a single CPU
Web server, is a software product designed to accelerate high-volume content transfers up to
700% over the Internet and Intranets.
What makes Acceloop interesting, for the purposes of this report, is that the company also
offers an optional technology under which content can be delivered on a peer-to-peer basis
across a network of user PCs. It calls this product the “Streaming Server Accelerator.”
The SSA is still in beta. The company says it can triple the number of concurrent users of live
streaming.
Financial Standing
Acceloop was seed-funded by private investors.
Current Situation
Acceloop is at work developing versions of its core Web Server Accelerator for Apache and
iPlanet Web servers, which should be ready before the end of this year.
Net/Net

It’s too soon to tell what kind of traction this new company will get in the marketplace. The
notion of accelerating content delivery from the Web server is sensible in today’s current
architectures, but certainly does little if anything to alleviate the underlying infrastructure cost
burden.
The addition of the peer --to-peer streaming functionality, however, intrigues us. Perhaps it will
be through Acceloop -- or another company like it -- that traditional infrastructure-based
CDNs can inconspicuously begin to experiment with PC-chained content delivery at the edge of
their infrastructure-based network.
If they do, Acceloop has quite a promising future ahead in the CDN world.

AllCast
535 West 34th St.
New York, NY 10001
Phone: (646) 792-2435
www.allcast.com
History
Founded in 1996, AllCast execs claim bragging rights as having the earliest patent in the P2P
CDN space. The company was awarded a patent for its “peer-to-multi-peer” audio
distribution software in 1999.
Services
AllCast’s software, the AllCast Broadcaster, is licensed to content creators on a per-listener
basis. The license starts at a 100 user license for one year at $500 and goes up from there.
AllCast is also willing to do event-based pricing or work with other fee models.
The company is proposing to offer dramatic savings over the infrastructure-based players.
For example, if a radio station incurred an $11K per month bill for audio Webcasting (based on
20 KB/second x 86,400 seconds per day x 30 days per month x 200 average listeners),
AllCast says it could bring costs down to roughly $200 per month --including a high-speed
connection. Its fees are based on a MB transfer cost of 0.009 cents/MB.
Financial Standing
AllCast is privately funded and does not provide additional information on its funding sources.
Current Standing
Allcast began beta testing of its technology early this year, offering free access to the beta
version of the broadcasting software. At the same time, users began to download the plug-in
that makes their Webcasts available on Windows Media Player -- a total of 27,000 users as of
mid-August, the company said.
In September, AllCast made additional progress in getting its product established, announcing
that it would be providing the software behind BRS Media’s My Radio.FM online broadcasting
service and professional AM/FM division WEBCASTi (http://www.webcasti.fm). The same
month, AllCast announced that it was chosen to

Webcast independent filmmaker group Independent Feature Project’s Market Seminars and
Panels taking place in late September/early October.
Meanwhile, in a step designed to underscore its legitimacy as a broadcaster, AllCast struck a
deal with Internet broadcasting measurement service MeasureCast in October. Under the
terms of the deal, MeasureCast will monitorr audience size and demographics for AllCast’s
broadcasts.
Net/Net
With roots extending back to 1996, AllCast has had time to finesse its product and its pricing
structure carefully. Working its technology as a plug-in to Windows Media Player certainly
doesn’t hurt on the consumer acceptance end.
At the same time, its current pricing structure is sensible and solid, with a relatively simple
structure and an entry-level price point accessible to smaller operators.
Still, at this stage, it appears that AllCast will be fighting a savage competitive battle for audio-
only delivery with ChainCast, one in which marketing, M&A dollars and a dash of plain luck
will play a role.
Ultimately, we don’t anticipate that AllCast and ChainCast will continue to compete in the same
niche for more than six or eight months. Instead, we predict that will be acquired by a
broadcasting or media company -- or even a telco such as AT&T with media interests -- that
wants to control the Internet media side of their business.

Blue Falcon Networks
The Bradbury Building
304 S. Broadway, Suite 596
Los Angeles, CA 90013
Phone: (213) 617-6900
www.bluefalcon.com
History
Blue Falcon was founded in 1995. Previously, the company was known as Static and focused
on online gaming technology, particularly P2P-related distributed networking. Over the past
three years, the company has been transforming that technology into the P2P content delivery
platform now moving into production. The company renamed itself Blue Falcon recently and
dedicated itself publicly to the streaming business.
Services
Blue Falcon distinguishes itself as a company offering streaming audio and video services that
work when delivered to large-scale audiences. One service, Streamer Live, is designed
deliver real-time Webcasts. Streamer On Demand, meanwhile, provides on-demand
streaming content (for example, when customers find and choose to play a particular video at
their leisure).
The company offers its content creator/broadcaster customers two ways to buy its streaming
technology.
Choice one is licensing. Blue Falcon’s first, flagship broadcast customer, Radio Free Virgin, is
leasing Blue Falcon’s technology and managing that technology at its own collocation facility.
Blue Falcon charges a per-megabytes-transferred licensing fee (similar in spirit to the per-CPU
fees charged by Oracle for its databases).
Blue Falcon also offers a hosted service in which it hosts its own application and then charges a
different per-megabyte fee for streaming data transferred.
Financial Standing
Blue Falcon is closemouthed about its financing. The company is funded by VC firm Zone
Ventures, but execs don’t disclose how much it has received, considering the numbers to be
proprietary. As of late October, the company expected to close soon on a second funding
round.

Current Situation
Blue Falcon continues to take steps to reinforce its position as a credidble Internet media entity.
Most recently (as of this writing), it signed a deal with Internet broadcasting measurement
service MeasureCast, a third-party streaming audience measurement service. The measurement
deal brings Blue Falcon in line with non-P2P Internet broadcasting companies, which have
gradually begun to accept the necessity for media monitoring techniques comparable to those
employed by offline broadcasters.
As of late October, the Radio Free Virgin deployment was scheduled to roll out to 44 channels,
available to the 3 million users who had downloaded Virgin’s player. In the works were deals
with enterprise customers to bring up enterprise streaming projects, in what Blue Falcon
executives describe as a wave of post-terrorism nervousness over physical travel.
Net/Net
There’s plenty to say about this company, which appears to have a technically robust offering
on the way as well as having snagged a high-profile customer. However, Blue Falcon still
hasn’t showed its hand entirely.
For the moment, we can only say that the company’s initial steps appear be well-taken. Let’s
see if Blue Falcon can build its initial, visible success with Radio Free Virgin into a few
entertainment players or wedge open the enterprise streaming door.

CenterSpan Communications Corp.
7175 NW Evergreen Pkwy, Suite 400
Hillsboro, OR 97124
Phone: (503) 615-3200
www.centerspan.com
History
CenterSpan evolved from a game controls company called ThrustMaster Corp. In September
1999, the company dumped its hardware business along with the ThrustMaster name and
turned to P2P technologies.
In December 2000, a key turning point in its development, CenterSpan acquired the assets of
Scour Inc., the company behind consumer online P2P service Scour. (The engineers behind
Scour have since moved on to form Red Swoosh. See company profile, below.)
More recently, in August of this year, CenterSpan acquired a group of technical assets from
Supertracks Inc., including the company’s BridgePort Music System, a secure local cache
streaming technology, along with its digital music distribution technology and several associated
patents-in-process.
The beta version of Scour’s retooled version launched in March of this year. A commercial
edition should launch before the end of 2001.
Services
CenterSpan’s C-star application allows content owners to deliver content via downloadable
files or via streaming media formats on a peer to peer basis. CenterSpan’s research has
concluded that their service can be delivered at 1/3 the cost of client-server driven downloads
and 1/9 the cost of existing streaming media technologies.
Its C-star service can be deployed either as part of a content creator’s existing infrastructure or
used as a turn-key product with custom templates.
The templates provide customer-specific views into a given content creator’s catalog. With
these tools, the idea is to allow customers to create their own branded content catalogs, and
then control the access rights of each individual subscriber.
CenterSpan’s service is designed to offer the maximum possible Digital Rights Management
controls. Currently using Microsoft’s Windows Media DRM technology, the C-star platform

compresses and encrypts content in a DRM wrapper, divided into segments, divided further
into packets and sent across the network, themselves encrypted. Packets are later assembled
on the desktop, as taken disparate user PCs with copies, by users with permission to receive
the content.
Financial Standing
Even in a down economy, there are advantages to being a public company. Despite slumping
tech stock prices, CenterSpan was able to finance its recent purchase of Supertracks assets
with approximately 67,000 shares of stock (valued at $750K) rather than a pure cash outlay.
This puts it ahead of other P2P players, most of whom are private and will almost certainly need
to invest a meaningful amount of cash if they want to do mergers or acquisitions.
As of June 30, 2001, according to its most recent quarterly filing, CenterSpan had $11.4 million
in working capital, a better financial position than many Internet startups. (Despite its age as a
company, its business model is new and that makes it a startup in our book.) Flows of cash
have come in periodically during the year, with investments from a varied group of players.
Sales revenue is still more or less non-existent, however. It remains to be seen how the Scour
launch will do in bringing the company up to profitability.
Current Situation
All told, CenterSpan seems to have set itself up nicely to begin to sell. The question now, which
we haven’t been able to answer to our satisfaction, is whether it can ramp up income quickly
enough to recoup all of this expense and survive solely on its own revenues.
While free Internet file sharing services have been quite popular, it’s still a very tricky open
question whether people are willing to pay for downloadable or streaming content. Advertiser-
supported models are familiar from the broadcast media, but pay-for-play is still a bit new.
Net/Net
We’re calling this one a wait-and-see playÖ.almost a coin flip .
CenterSpan is at a critical point in its evolution, having acquired or developed the technologies it
needed and gone through beta with a popular platform (Scour). The rubber, so to speak, is
about to hit the road.
If demand for paid-for digital content is there, we think CenterSpan’s in an excellent position to
meet it, but we’re still not sure entertainment buyers are ready.

ChainCast Networks
100 Century Center Court, Suite 800
San Jose, CA 95112
Phone: (408) 467-4040
www.chaincast.com
History
ChainCast was founded in March 1999, and took in its first equity funding in December of that
year. The company holds 10 patents related to various aspects of its “chaincasting” technology.
Services
ChainCast is offering managed P2P content delivery services and at the same time, pitching its
products to network service providers and telcos. Pricing models for ChainCast’s managed
services are listed below:
ChainCast pricing models
95th Percentile Bandwidth (Mbps)
The 95th percentile bandwidth model is based on the average of samples of the bandwidth
served (in megabits) during each 5-minute segment of the month. The top 5% of the 5-minute
averages are discarded, and the Webcaster is charged for the next highest, or peak 5-minute
average.
Aggregate Monthly Volume (GB or TB)
The aggregate volume served model is based on the total number of bytes delivered over a
monthly period, which is a function of the number of user sessions, the length of the sessions,
and the bit stream encoding rate.
Monthly Average Bandwidth (Mbps)
The monthly average bandwidth in Mbps is the aggregate volume delivered during the month
(converted from bytes to bits) divided by the number of seconds in the month or alternately, the
average of all of the 5-minute segment averages for the month.
Information courtesy of ChainCast Networks

Financial Standing
To date, ChainCast has taken in over $15 million in venture funding. Its backers include
Bellwether Investments, Creative Technology Ltd., Cuscaden Investments, MediaCorp
Interactive, Mid-City Investments Holding, WIIG Global Ventures and the founders of Creative
Labs and InnoMedia Pte Ltd.
Current Situation
ChainCast has been conducting ongoing beta tests of its audio Webcasting technology since
mid-2000 with six radio stations. Over that time, of course, it has been tweaking its product to
ready it for wider adoption.
The company has now begun tests with telcos, but is not yet prepared to go public with the
results of those tests. It is also hosting an as-yet unnamed event for a media firm, an event which
expected to draw as many as 100,000 listeners.
ChainCast has not announced any post-beta commercial deals as of yet, though talks with
company officials suggest that they may do so shortly if the trials above go well.
However, ChainCast did reach a landmark with its acquisition of Internet radio streaming and
ad replacement vendor StreamAudio. By acquiring StreamAudio, ChainCast went overnight
from a technology vendor to a large-scale distributor, an important distinction in a tightening
marketplace.
Net/Net
ChainCast is one of only a few players (alongside contenders such as Allcast, Blue Falcon
Networks and vTrails) proposing to deliver high-quality audio content in real time via a P2P
infrastructure on the public Internet. The competition between these companies is going to be
very intense over the next six to twelve months. With its acquisition of StreamAudio, ChainCast
has done something decisive to stand out; it will be interesting to see how it leverages this asset
competitively.
Looking at ChainCast’s broad base of VC support, we believe the company has reason for
financial enthusiasm, though it is worth noting that the original cash infusion came before the Net
biz crash. If ChainCast is meeting whatever goals the original VCs had set, there may well be
more money to come -- and if the money goes into more strategic acquisitions, the game could
get very interesting indeed.

We’re also a fan of its managed services/telco channel sales model; we believe these are likely
to appeal to the broadest possible audience while keeping good control of reputation-building
quality and results reporting. For enterprises, managed P2P content delivery services also bear
a close resemblance to other network services they’re used to buying, and that can only help to
close deals.
Focused as it is only on audio, the company takes some flack from its competitors with video
capacity, who view the ChainCast product as too limited. For our part, we believe this
purported limitation could be a strength; after all, video and audio media markets are discrete
enough that focusing on just one may lead to a better strategic focus. Still, that may change as
the market matures, and this is a risk factor for this otherwise well-positioned firm.

EverNet Systems Inc.
225 Velarde Street
Mountain View, CA 94041
Phone: (650) 386-5545
www.evernet.com
History
Publicly launched in August of this year, EverNet has been in stealth mode for about two years.
The company’s founder and current acting chief executive officer, Mike Helm, previously ran an
ISP, as well as having developed and operated several Web sites. Many of the technical staff
members on the engineering team had previously worked together at desktop entertainment
software company Sierra On-Line of Bellevue, Wa.
Services
To begin taking advantage of EverNet’s service, a user clicks on a link connected to an
EverNet supported file. The company’s thin client installs automatically on a user’s desktop the
first time the user requests a file. The client is currently about 800K in size; EverNet is working
to reduce it to 500K.
After connecting, the parent EverNet server determines which 20 connected desktops are
closest to that user within 20 milliseconds. Each of these 20 desktop computers sends the user
little pieces of the requested file. If a sending desktop goes offline, gets overloaded or begins
sending a corrupted file, EverNet’s servers automatically switch the feed off to another
computer. According to EverNet’s research, file delivery performance improves as users join
the party.
EverNet’s technology is designed to work with all standard digital content file types, including
Windows Media, RealPlayer and QuickTime.
Standard pricing for services starts at approximately 2/10 of a cent per megabyte, or about $2
per gigabyte. The company offers discounts for greater volume. The company expects to set a
minimum delivery level, as their P2P approach won’t work for tiny files.
Financial Standing
To date, EverNet has taken in $1.8 million in angel funding from unnamed investors. However,
the company has invested all of that money in development of its proprietary technology and
sales activity. Other than Filmspeed (see below), the company has not publicly identified any

live customers; if it doesn’t have any, it’s likely that EverNet urgently needs more customers or
customers to stay afloat.
Current Situation
At present, EverNet is just beginning to build and audience and connect with partners.
Its first content partner relationship is with Filmspeed, which distributes DVD-quality motion
picture content to computers and mobile devices. The two companies have worked together to
create a joint product, Filmspeed RipTide powered by EverNet, which can be tested at Error!
Bookmark not defined..
Net/Net
EverNet appears to be going head to head with Kontiki, CenterSpan and perhaps Uprizer.
This is a race it will lose unless a) execs can demonstrate a compelling cost savings, which to be
fair is entirely possible or b) the company offers services so compellingly differentiated that
nobody cares about, say, the sexiness of a company backed by Marc Andreessen and Jim
Barksdale.
We have talked with EverNet and think the players there are shrewd enough to make it happen,
at least if their resources hold out. We don’t expect to see them hold on unless they cut a few
more deals quickly -- either VC or customers, though customer deals have more lasting value --
but if they can get cash flowing we see them as a real contender.
One remaining question, which will dog Kontiki, CenterSpan, EverNet and any other future
download-oriented P2P content delivery player, is whether users will be willing to wait for
downloads in order to experience high-res movies and the like. That is simply an unknown at
the moment. We’re skeptical about this -- the Internet is an “instant” culture -- but feel it’s too
soon to call the play. Our guess is that corporate markets delivering say, training films may take
to this better than impatient entertainment masses but time will tell.

Kontiki
1350 Villa Street, Suite 200
Mountain View, CA 94041
Phone: (650) 625-3000
www.kontiki.com
History
Kontiki began life as an ultra-stealthy firm -- the mysterious Zodiac Networks -- then burst into
life in August, splashing its new name about and garnering lots of press attention in the process.
The attention was probably due in substantial part to the leadership of Mike Homer, a former
executive vice president with Netscape Communications Corp, and investment from Netscape
whiz kid Marc Andreessen and leader Jim Barksdale. Thanks to the presence of Homer and
other former Netscape players, Kontiki snared prestige capital early on and has gotten a jump
on its plans to deliver audio, video, software and digital media.
The company initially described its technology as peer to peer at the outset but has since
backed away from the P2P label, according to press representatives, who declined to be
interviewed for our research. Still, we think Kontiki belongs in this report, and so we have still
followed through with our profile here.
Services
Kontiki’s core service is a “delivery network” leveraging the power of user PCs to cut media
delivery costs. Kontiki files are sent from caches held at the edge of its delivery network. To
avoid wasting bandwidth, the system sends files at off-hours when connections are typically
underutilized.
To dispatch the files, Kontiki uses what it calls “Adaptive Rate Multiserving”, connecting user
computers to many servers in a network simultaneously. To receive a file, a Kontiki-enabled
computer requests small parts of a file from each upstream computer. The system adjusts the
serving pattern based on the available bandwidth and throughput. In this manner, user
computers pass files on to user computer computers down the line.
Among the core features of the system are digital rights management controls which prevent
content from being played without permission of its creators.

While Kontiki declined to speak with us, CEO Mike Homer was quoted in Interactive Week
magazine as saying that Kontiki is planning to charge customers a $10K activation fee and a
minimum of $6K per month to deliver 500,000 MB of data.
Financial Standing
Kontiki has taken in $18 million in its first round of funding. Investors include some of the more
prestigious VC money out there, including the Barksdale Group and Benchmark Capital. It has
also attracted funding from several prominent angel investors, including Marc Andreessen and
VeriSign chief executive officer Stratton Sclavos. Given the high profile of Kontiki CEO Homer
and other ex-Netscape players in the company, odds seem high that Kontiki can attract
additional funding despite a currently hostile environment for venture investment.
Current Situation
Kontiki is positioning itself to compete most heavily in a sector -- time-shifted media delivery --
which has attracted fewer competitors to date. While it offers both streaming and on-demand
delivery of media sources, most coverage of Kontiki has emphasized its on-demand service,
and in fact, on-demand service does differentiate it from the pack more than streaming services
do anyway.
Though a meaningful and growing number of players are making similar sounding offers to
distribute streaming media, only one other company on our radar screen -- EverNet -- is
focused tightly on the time-shifted delivery of the P2P content delivery market. While there may
very well be others, there don’t appear to be many.
For this reason, we believe that Kontiki is currently in an excellent position to test out its theory
(and we see it as a theory rather than a known fact) that consumers and businesses want to
download files at off times then view them at their leisure from files stored on their hard drive.
However, they’re also in a good position to fall on their if this idea doesn’t catch on quickly. As
we see it, their period of opportunity is relatively narrow.
Net/Net
If fame and prestige mean something -- and they probably do in the corporate market --
Kontiki already has something of a pole position in the time-shifted peer to peer market. On the
other hand, this is still an extremely volatile space, and market frenzy isn’t enough to sustain a
company’s lead more than six months at best.
We think corporate buyers seeking to distribute training materials will be somewhat interested in
Kontiki, but not necessarily that much more than streaming players such as ChainCast, AllCast

and vTrails. Enterprises with a strong outsourcing philosophy might like Kontiki’s outsourced
approach, but prefer to access vTrails’ rack-mounted hardware solution through an ISP to
parcel out bits when they deliver the CEO’s speech across the WAN.
On the entertainment and media side of things, meanwhile, we think the “sexiness quotient” will
go a lot further. The mere glamour of techno-plutocrats like Andreessen and Homer, who are
quite the counterparts of the moguls in the entertainment business, will resonate well with their
opposite numbers in the music, film, television and movie business. In this sector, we think
they’ll be hard to beat, at least if the “schedule download, play later” model becomes popular
with consumers.

Red Swoosh, Inc.
1100 Glendon Ave., Ste. 1700
Los Angeles, CA 90024
www.redswoosh.com
History
Red Swoosh has been in extreme stealth mode since its launch. (The only telephone number
listed for the company at the time of this writing was disconnected!)
All that we know about RedSwoosh to date is that its principals were engineers behind Scour
Inc.’s multimedia search technology prior to the buyout of Scour’s assets by CenterSpan. Its
principals have been quoted in the press claiming that they have mature technology on par with
if not in advance of P2P content delivery competitors.
Services
Because RedSwoosh has not responded to our inquiries to date, we can only quote the very
brief self-description on their site to describe their services:
“Red Swoosh builds software that has knowledge of all caches of data objects on the global
network and coordinates the most cost effective delivery and storage of those objects at a
quality of service that is unattainable on today's network.”
Financial Standing
No information available to date.
Current Standing
RedSwoosh began a content delivery beta program in September 2001, and according to its
site, hopes to become the premier provider of next-generation, intelligent digital media
delivery systems to enterprises, media companies and ISPs. In other words, it’s clearly not
going for the end user audience pursued by some of its P2P CDN competitors profiled above in
this report .
Net/Net
From the little information the company has provided on its site, RedSwoosh seems to be
positioning itself as a player in the space shared by Speedera and other networking players,

making their bread optimizing networks for content delivery.

Uprizer Inc.
1007 Montana Ave., Suite 323
Santa Monica, CA 90403
www.uprizer.com
History
Founded in 2000 but debuting publicly in October 2001, Uprizer has proven to be one of the
stealthier firms on this list. Among other things, its phone number is nowhere to be found on its
site or the Web.
Uprizer was in deep cloaked mode for quite some time, though an April 2001 announcement on
Intel’s site noted that Intel and a few prominent venture capital players had invested in the new
firm.
The firm comes with a healthy P2P pedigree. Uprizer was co-founded by Ian Clarke, who is
already somewhat of a P2P legend as the creator of the million-plus node Freenet file-sharing
platform.
Services
Uprizer’s core offering is the KARMA (Key Accessed Redundant Memory Architecture)
Platform, a P2P content distribution app with distinct grid computing features. It is capable of
operating solely with client software, available for a wide range of computing devices, but a
server is available for companies that wish to license one for redundant content availability.
In addition to content distribution, Uprizer aggregates unused storage capacity across these
devices, and presents this storage capacity in the form of a virtual disk it calls the KARMA
Drive.
KARMA is available for license, or as a managed service (complete with Service Level
Agreement) through an Uprizer certified Service Provider. Uprizer is also looking into
distribution deals of various kinds, including providing branded KARMA clients to content
providers and preloaded clients for PC manufacturers.
Along the way, Uprizer expects to save content producers money. The company claims
KARMA can decrease content distribution costs by more than two-thirds, from an average
$1.50/GB to less than $0.50/GB.
Financial Standing

Uprizer raised $4 million this year from a group of well-heeled investors including Intel Capital,
Kline Hawkes Pacific LP and Shugart Ventures. The company does not appear to have any
commercial customers as of yet, so it’s unclear how they are situated other than having VC
funding. With the money coming in after much of the costly initial development was completed,
however, Uprizer can plow a good percentage of the funding into sales. That bodes well for
getting their message across despite a crowded marketplace.
Current Situation
As of October 2001, Uprizer was beta testing its platform, doing scalability tests as part of a
testbed agreement with IBM. Working in concert, the computing giant and Uprizer are
simulating the effects of various forms of connectivity (e.g. cable modems, DSL connections and
dial-up) on file distribution performance.
Executives expect to launch KARMA commercially in the first quarter of 2002. Going forward,
Uprizer is exploring the possibility of turning its client into a chipset for use in wireless phones.
Uprizer is also preparing to offer a streaming P2P distribution product sometime in the middle of
2002.
Net/Net
Uprizer is coming into the P2P distribution market at a good time, as interest is rising and
traditional CDNs weakening. What’s more, Uprizer’s hybrid architecture -- with aspects of
grid computing and content distribution to offer -- should offer extra value to enterprise users
hoping to experiment with varied P2P computing modes.
Uprizer is investing considerable energy in the wireless space, and this could distract executives
from plucking the low-hanging fruit of corporate streaming services which we see as a clear
option for most players. On the whole, however, Uprizer’s approach seems smart and their
security infrastructure well-considered, something which may get its salespeople in the door
where others fail.

vTrails
65 Broadway, Ste. 902
New York, 10006 NY
Phone: (212) 742-1395
www.vtrails.com
History
VTrails was founded in 1999 with funding from an Israeli venture capital firm. Like many Israeli
firms with aggressive expansion plans, the company maintains an R&D facility in Tel Aviv but
located its sales, marketing and media functions in a New York city office.
Services
Unlike many of its competitors, vTrails is selling hardware rather than software services.
VTrails systems vary not only in price but also in how many channels they can serve. At
present, all models except the vTrails Basic are multichannels and can concurrently serve six to
10 broadcasts. The Basic can serve two audio channels.
The vTrails system sits on a rack next a company or ISP’s media server, and works with
standard media servers including Real, QuickTime and Windows as well as ad insertion
technologies such as Vignette.
System prices are structured as follows (as of August 2001):
Financial Standing
VTrails received seed funding from Israeli incubator teamDCL in August of 1999. The
company has not disclosed any other financial information.

Current Situation
vTrails appears to be in early startup mode. The company has begun beta testing with Chicago-
based Webcasting firm RadioWave and San Francisco-based Internet broadcaster
CreativeBroadcasting, and recently gave its technology a public debut at the Streaming Media
West trade show, but has not begun a formal product rollout.
Net/Net
vTrails is unique in its hardware-based approach. Depending on how easily -- or painfully -- it
installs in server racks, this could be a breakthrough technology for network service providers,
simply because it fits in with the business model and approach providers expect. However, we
know too little about the company’s financial status and administrative capabilities to know
whether it can execute on sales and manufacturing well enough to capitalize on that promise.

P2PContent

More Related Content

What's hot (20)

Viewers also liked (6)

Similar to P2PContent (20)

P2PContent