Buildvpn1.pdf

Part
The VPN Overview
1
VPN1 6/9/03 6:00 PM Page 1

Chapter
VPN-in-Brief
3
1
1.1 VPN Overview
This is the information age. We no longer have to commute physically from one
place to another to complete a set of tasks or to gather pieces of information.
Everything can be done virtually with a mouse click on an online host. In a way,
everything we do in our daily lives is related in one way or another to infor-
mation access. This has made information sharing almost mandatory and indis-
pensable. These days, a customer can retrieve and compare products or services
information promptly online, anytime, anywhere. For competitive reasons,
organizations that provide this information have to make the information
readily available online. In other words, the concept of a shared infrastructure
is undisputedly important. A shared infrastructure is none other than a public
network. At present, the biggest public network is the Internet, which has over
100,000 routes and is still growing rapidly.
As more and more companies link up their corporate network to the Inter-
net, we are faced with an inevitable issue—information security. Sharing infor-
mation on a public network also implies giving access and visibility to everyone
who wants to retrieve these data. What if the person who has the accessibility
and visibility to the information decides to create havoc? Some of the general
threat types that are posed by malicious hackers include eavesdropping, denial
of service, unauthorized access, data manipulation, masquerade, session replay,
and session hijacking.
How do we ensure the safe passage of data across a shared infrastructure?
The answer is to deploy a secured virtual private network (VPN). VPNs are net-
works deployed on a public network infrastructure that utilize the same secu-
rity, management, and quality of service policies that are applied in a private
network. VPNs provide an alternative to building a private network for site-to-
site communication over a public network or the Internet. Because they operate
across a shared infrastructure rather than a private network, companies can
cost effectively extend the corporate WAN to telecommuters, mobile users, and
VPN1 6/9/03 6:00 PM Page 3

remote offices as well as to new constituencies, such as customers, suppliers,
and business partners.
Traditional private WANs connect customer sites via dedicated point-to-point
links. This means that multiple independent circuits have to terminate at the
corporate network egress, making the deployment nonscalable and difficult to
maintain. VPNs extend the classic WAN by replacing the physical point-to-point
links with logical point-to-point links sharing a common infrastructure, allow-
ing all the traffic to be aggregated into a single physical connection. This
scenario results in potential bandwidth and cost savings at the network egress.
Because customers no longer need to maintain a private network, and because
a VPN itself is cheaper to own and offers significant cost savings over private
WANs, operation costs are reduced.
VPNs provide an alternative WAN infrastructure that can replace or augment
commercial private networks that use leased-line or frame relay/ATM networks.
There are two ways business customers can implement and manage their VPNs.
They can either roll out their own VPNs and manage them internally, or out-
source the VPN management to their service providers for a total VPN package
that is tailored to their particular business needs.
Last but not least, from the service providers’ perspective, VPNs are a fun-
damental building block in delivering new value-added services that benefit
their business customers as well as themselves. In this instance, the service
providers deploy the VPNs for their customers, and the customers need only
subscribe to the service providers for the VPN services.
1.2 VPN Types and Solutions
In this section we address three types of VPNs: remote access, site-to-site, and
firewall-based (a site-to-site variation). The variation between remote access
and site-to-site VPNs will become more ambiguous as new devices such as hard-
ware VPN clients, become more prevalent. These appear as a single device
accessing the network, albeit there may be a network with several devices
behind it. In all cases, the VPN comprises two endpoints that may be repre-
sented by routers, firewalls, client workstations, or servers.
1.2.1 Remote access
Remote access VPNs or virtual private dialup network (VPDN) are deployed
for individual remote users, commonly referred to as mobile users and telecom-
muters. In the past, corporations supported these remote users via dialup net-
works, which required the remote users to make toll calls to access the corporate
network directly. This was not a cost-effective solution, especially when an inter-
national user made a call back.
With the introduction of remote access VPNs, a mobile user can make a local
call to their Internet service provider (ISP) to access the corporate network with
their PC via the Internet wherever they may be. Remote access VPNs are an
4 The VPN Overview
VPN1 6/9/03 6:00 PM Page 4

extension of the traditional dialup networks. In this case, the software on the
PC provides a secure connection, often known as a tunnel, back to the corpo-
ration. Since the users need only make local calls, the operation cost is reduced.
Remote access VPNs and their corresponding technologies are beyond the scope
of this book.
1.2.2 Site-to-site
Site-to-site VPNs are deployed for interconnecting corporate sites. In other
words, the network of one location (site) is connected to the network of another
location (site) via a VPN. In the past, a leased line or frame relay connection
was required to connect the sites; however, these days, most corporations have
Internet access. With Internet access, leased lines and frame relay circuits can
be replaced with site-to-site VPNs. Site-to-site VPNs are an extension of legacy
WAN networks.
Site-to-site VPNs can be further viewed as intranet VPNs or extranet VPNs.
Intranet VPNs refer to connections between sites that all belong to the same
organization. User access between these sites is less restraining than for
extranet VPNs. Extranet VPNs refer to connections between an organization
and its business partners. User access between these sites should be tightly con-
trolled by both entities at their respective sites.
1.2.3 Firewall-based
A firewall-based VPN is intrinsically a site-to-site implementation. Firewall-
based VPN solutions are not a technical but a security issue. They are deployed
when a corporation requires more advanced perimeter security measures for
its VPNs. Corporations can enhance their existing firewalls to support firewall-
based VPNs.
1.3 VPN Terminology
This section lists some of the common VPN terminology that is used in the sub-
sequent sections:
Provider network (P-Network): the service provider infrastructure that is
used to provide VPN services.
Customer network (C-Network): the part of the network that is still under
customer control.
Customer site: a contiguous part of the C-Network that can comprise many
physical locations.
Provider (P) device: the device in the P-Network with no customer connec-
tivity and without any “knowledge” of the VPN. This device is usually a
router and is commonly referred as the P router.
VPN-in-Brief 5
VPN1 6/9/03 6:00 PM Page 5

Provider edge (PE) device: the device in the P-Network to which the CE
devices are connected. This device is usually a router and is often referred as
the PE router.
Customer edge (CE) device: the device in the C-network that links into the
P-network; also known as customer premises equipment (CPE). This device
is usually a router and is normally referred as the CE router.
Virtual circuit (VC): logical point-to-point link that is established across a
shared layer-2 infrastructure.
1.4 VPN Models
VPN services can be offered based on two major paradigms:
1. The overlay VPNs, whereby the service provider furnishes virtual point-to-
point links between customer sites.
2. The peer-to-peer VPNs, whereby the service provider participates in cus-
tomer routing.
In the following sections, we discuss these two different models in details.
1.4.1 Overlay model
The overlay VPN is deployed via private trunks across a service provider’s
shared infrastructure. These VPNs can be implemented at layer-1 using
leased/dialup lines, at layer-2 using X.25/frame relay/ATM Virtual Circuits, or
at layer-3 using IP (GRE) tunneling.
In the overlay VPN model, the service provider network is a connection of
point-to-point links or virtual circuits (VCs). Routing within the customer
network is transparent to the service provider network, and routing protocols
run directly between customer routers. The service provider has no knowledge
of the customer routes and is simply responsible for providing point-to-point
transport of data between the customer sites.
Figure 1.1 illustrates the deployment of an overlay VPN. The scenario adopts
a hub-and-spoke topology whereby the Paris site is the hub, and both the
London and Zurich sites are the spokes. The London site is linked up to the
Paris site via a point-to-point VC #1. Likewise, the Zurich site is linked up to
the Paris site via a point-to-point VC #2. In this instance, the layer-3 routing
adjacencies are established between the CE routers at the various customer
sites, and the service provider is not aware of this routing information at all.
As illustrated in Figure 1.2, from the perspective of the CE routers, the service
provider infrastructure appears as point-to-point links between Paris–London
and Paris–Zurich.
The overlay VPN model has two further constraints. One is the high level of
difficulty in sizing the intersite circuit capacities. The other is the requirement
6 The VPN Overview
VPN1 6/9/03 6:00 PM Page 6

of a fully meshed deployment of point-to-point links or VCs over the service
provider’s backbone to attain optimal routing.
1.4.1.1 Layer-1 implementation. Figure 1.3 illustrates the overlay VPN layer-1
implementation, which adopts the traditional time division multiplexing (TDM)
solution. In this scenario, the service provider assigns bit pipes and establishes
the physical-layer (Layer-1) connectivity between customer sites via ISDN, DS0,
T1, E1, SONET, or SDH, and the customer is accountable for implementation
of all higher layers, such as PPP, HDLC, and IP.
VPN-in-Brief 7
Service Provider Network
Customer Site
Customer Site
PE Device
PE Device
PE Device
VC #2
VC #1
London-CE
Zurich-CE
Paris-CE
Customer Site
Layer-3 Routing Adjacency
Layer-3 Routing Adjacency
Figure 1.1: The overlay VPN model.
Paris-CE
London–CE Zurich–CE
“Point-to-point link” “Point-to-point link”
Figure 1.2: Perception of the SP infrastructure from the CE routers.
VPN1 6/9/03 6:00 PM Page 7

implementation, which adopts the traditional switched WAN solution. In this
scenario, the service provider is responsible for establishing layer-2 VCs
between customer sites via X.25, Frame Relay, or ATM, and the customer is
accountable for the IP layer and above.
implementation, whereby the VPN is implemented with point-to-point IP-over-
IP tunnels. This is commonly referred as IP tunneling whereby a destination
8 The VPN Overview
IP
SONET, SDH
DS0, T1, E1
ISDN
HDLC
PPP
Figure 1.3: Overlay VPN layer-1
implementation.
IP
ATM
Frame Relay
X.25
implementation.
IP
Generic Routing
Encapsulation
(GRE)
IP Security (IPSec)
IP
implementation.
VPN1 6/9/03 6:00 PM Page 8

can be reached transparently without the source having to know the topology
specifics. Therefore, virtual networks can be created by tying otherwise uncon-
nected devices or hosts together through a tunnel. Tunnels also enable the use
of private network addressing across a service provider’s backbone without
the need for network address translation (NAT). Tunnels are established with
generic routing encapsulation (GRE) or IP security (IPSec). The GRE imple-
mentation is simpler and quicker but less secure, whereas the IPSec deploy-
ment is more complex and resource-intensive (CPU cycle) but it provides more
robust security.
GRE tunnels provide a specific conduit across the shared WAN and encapsu-
late traffic with new packet headers to ensure delivery to specific destinations.
Since traffic can enter a tunnel only at an endpoint, the network is private.
GRE tunnels do not provided true confidentiality but can carry encrypted
traffic. For better protection, GRE can be used in conjunction with IPSec.
When we deploy a layer-3 VPN over a public network, the VPN tunnels are
built across a distrusted shared infrastructure. IPSec provides robust security
services such as confidentiality, integrity, and authentication to ensure that sen-
sitive information is securely transported across these tunnels and not circum-
vented accidentally or intentionally.
IPSec is an Internet Engineering Task Force (IETF) standard that enables
encrypted communication between users and devices. It can be implemented
transparently and seamlessly into the network infrastructure. End users need
not have any knowledge that packets are being intercepted and transformed by
IPSec. Because it operates at the network layer (layer-3), IPSec is ideally posi-
tioned to enforce corporate network security. Since IPSec encryption works only
on IP unicast frames, it can be used in conjunction with GRE to alleviate this
deficit. This is because GRE is capable of handling the transportation of multi-
protocol and IP multicast traffic between two sites, which have only IP unicast
connectivity.
1.4.2 Peer-to-peer model
The peer-to-peer model adopts a simple routing scheme for the customer. Both
provider and customer network use the same network protocol and all the
customer routes are carried within the core network (service provider network).
The PE routers exchange routing information with the CE routers, and layer-
3 routing adjacencies are established between the CE and PE routers at each
site. Because peer-to-peer routing has been implemented, routing between sites
is now optimal. Fully meshed deployment of point-to-point links or VCs over
the service provider backbone is no longer applicable to attain optimal routing.
Since there is no overlay mesh to contend with, the addition of new sites is
easier, and circuit capacity sizing is not an issue. Because the service provider
now participates in customer routing, provider-assigned or public address space
needs to be deployed at the customer’s network, so private addressing is no
longer an option.
VPN-in-Brief 9
VPN1 6/9/03 6:00 PM Page 9

Figure 1.6 illustrates the deployment of a peer-to-peer VPN. In this scenario,
customer routing information is exchanged between Paris-CE and Paris-PE.
The customer routes are then broadcast through the core network to London-
PE and Zurich-PE, which in turn, propagate these routes to their respective CE
routers. The shared router and dedicated router approaches are derivatives of
the peer-to-peer model. We discuss these two approaches in Sections 1.4.2.1 and
1.4.2.2.
1.4.2.1 Shared PE router model. In this model, a common PE router that
carries customer routes is deployed. Individual customers routes are separated
with packet filters on PE-CE interfaces. The packet filters are managed so that
information goes to the proper site and different customers are separated. The
complexity of these packet filters results in high maintenance costs and a sig-
nificant impact on performance.
Figure 1.7 illustrates the Shared PE router model. In this scenario, there are
three separate VPNs: VPN-101, VPN-201, and VPN-301. These VPNs are
deployed over four different customer sites. VPN-101 is deployed for Paris as
well as Lyon; VPN-201 is deployed for Brussels; and VPN-301 is deployed for
Munich. In Figure 1.7, we can see that London-PE carries all the customer
routes for VPN-101, VPN-201, and VPN-301. Isolation between VPNs is
achieved with packet filters (access lists) on the PE-CE interfaces: Serial0/0,
Serial0/1, Serial0/2, and Serial0/3.
10 The VPN Overview
Customer Site
London-CE
Customer Site
Zurich-CE
Zurich-PE
London-PE
Routing information is exchanged
between CE and PE routers.
PE routers exchange customer
routes through the core network.
The customer routes are propagated to
other CE routers via the PE routers.
Customer Site
Paris-CE
Paris-PE
Layer-3 Routing
Adjacency
Layer-3 Routing
Adjacency
Layer-3 Routing
Adjacency
3
3
1
1
2
2
Figure 1.6: The peer-to-peer VPN model.
VPN1 6/9/03 6:00 PM Page 10

1.4.2.2 Dedicated PE router model. In this model, each customer has a
dedicated PE router that carries only its own routes. Customer segregation is
achieved through lack of routing information on the PE router. The P router
contains all customer routes and filters routing updates between different PE
routers using Border Gateway Protocol (BGP) Communities. Because each
customer has a dedicated PE router, this approach is expensive to deploy, and
hence it is not a cost-effective solution.
Figure 1.8 illustrates a dedicated PE router model. In this scenario, there are
two separate VPNs: VPN-101 and VPN-201. These VPNs are deployed over four
different customer sites. VPN-101 is deployed for two locations in Paris and one
location in Brussels, and VPN-201 is deployed for one location in London. In
Figure 1.8, we can see that the P router in the service provider network
contains all the customer routes for VPN-101 and VPN-201. It filters routing
updates between Paris-PE, London-PE, and Brussels-PE using BGP Commu-
nities. VPN-101 routes are tagged with a community value of 101:1, and VPN-
201 routes are tagged with a community value of 201:1. In other words, the P
VPN-in-Brief 11
VPN-101
Paris-CE
London-PE
London-PE Routing Table
VPN-101 routes
VPN-201 routes
VPN-301 routes
interface Serial0/0
description ** interface to VPN-101 customer
ip address 172.16.11.5 255.255.255.252
ip access-group VPN-101 in
ip access-group VPN-101 out
!
interface Serial0/1
ip address 172.16.11.9 255.255.255.252
!
interface Serial0/2
ip address 172.16.21.5 255.255.255.252
!
interface Serial0/3
ip address 172.16.31.5 255.255.255.252
PE router carries all
the customer routes.
Isolation between customers
is achieved with packet filters
on the PE-CE interfaces.
VPN-101
Lyon-CE
VPN-201
Brussels-CE
VPN-301
Munich-CE
Figure 1.7: Shared PE router model.
VPN1 6/9/03 6:00 PM Page 11

router will propagate routes with a community value of 101:1 to VPN-101 and
routes with a community value of 201:1 to VPN-201. Each customer or VPN
has a dedicated PE router that carries only its routes. Customer isolation is
simply achieved through the lack of routing information on the PE routers.
In this example, routes from Paris-CE1 and Paris-CE2, which are both in
VPN-101, are announced to a dedicated PE router—Paris-PE. Paris-PE in turn
advertises the routes to the P router. The P router uses a community list to
match routes with a community value of 101:1 and propagates the routes that
match this community value to another dedicated PE router, Brussels-PE,
which serves the customer site that is in VPN-101 at Brussels.
1.4.3 MPLS VPN model
MPLS VPN is a true peer-to-peer model that combines the best of both worlds.
It unites the customer security and segregation features implemented in the
overlay model with the simplified customer routing deployed in the traditional
peer-to-peer model. The MPLS VPN architecture is very similar to the dedi-
cated PE router model, except the dedicated per customer routers are imple-
mented as virtual routing tables within the PE router. In other words, customer
segregation is achieved through the concept of virtual routing and forwarding
(VRF) whereby the PE router is subdivided into virtual routers serving differ-
12 The VPN Overview
Paris-PE
London-PE
P Router
Each customer has a
dedicated PE router that
carries only its routes.
The P router contains all
customer routes.
Customer isolation is achieved
through lack of routing
information on the PE router.
Paris-CE2
London-CE
VPN-101
Paris-CE1
VPN-101
VPN-201
VPN-101 routes ONLY
router bgp 65000
neighbor 192.168.101.2 remote-as 65000
neighbor 192.168.101.2 route-reflector-client
neighbor 192.168.101.2 route-map VPN-101 out
!
route-map VPN-101 permit 10
match community-list 1
!
ip community-list 1 permit 101:1
!
P Router Routing Table
VPN-101 routes (community 101:1)
VPN-201 routes (community 201:1)
Brussels-PE
Brussels-CE
VPN-101
Figure 1.8: Dedicated PE router model.
VPN1 6/9/03 6:00 PM Page 12

ent VPNs (or customer sites). This establishes overlapping addresses in differ-
ent customer sites since each customer is now assigned an independent routing
table.
The PE routers hold the routing information only for directly connected
VPNs. As a result, the size of the PE routing table is significantly reduced. The
amount of routing information is proportional to the number of VPNs attached
to the PE router. As such, the PE routing table will still grow when the number
of directly connected VPNs increases. In addition, the PE routers participate in
customer routing, ensuring optimal routing between sites and easy provision-
ing. Full routing within the service provider backbone is no longer required
because multi protocol label switching (MPLS) (see Chapter 7 for more in-depth
MPLS concepts), and not traditional IP routing, is used to forward packets. The
details of the MPLS VPN architecture are discussed in Chapter 8.
Figure 1.9 illustrates the MPLS VPN model. There are two separate cus-
tomers—Customer X and Customer Y. Customer X spans three different sites,
while Customer Y has only one site. Customer isolation is achieved with a dedi-
cated per customer virtual router. In Figure 1.9, we can see that routing within
the PE router is split into two separate planes, one for VPN routing and the
other for global IP routing.
In the VPN routing plane, the PE router is subdivided into virtual router
X which serves customer X, and virtual router Y which serves customer Y.
Each of these virtual routers participates in customer routing and keeps
an independent VPN routing table for its respective customers. Meanwhile, the
PE router also has a global IP router in the global IP routing plane that take
cares of the IGP and non-VPN BGP routing between the various PE and P
routers.
VPN-in-Brief 13
Customer X
Site #1
Customer X
Site #2
Customer Y
Site #1
Customer X
Site #3
Virtual Router X
VPN Routing Table for
Customer X
Virtual Router Y
VPN Routing Table for
Customer Y
Global IP Router
Global IP
Routing Table
IGP & non-
VPN BGP
P Router
PE Router
Figure 1.9: MPLS VPN model.
VPN1 6/9/03 6:00 PM Page 13

1.5 Comparison Between Various VPN Implementations
This section compares some of the benefits and drawbacks between overlay
VPN and peer-to-peer VPN. Table 1.1 gives a quick appraisal of the upside and
downside of the overlay VPN model versus the peer-to-peer VPN model.
Table 1.2 illustrates the benefits of the MPLS VPN model, which are com-
bined from the overlay and peer-to-peer VPN models for the best of both worlds.
1.6 IPSec versus MPLS VPNs
Both IPSec and MPLS VPNs are implemented at the network layer of the OSI
model. In other words, they are layer-3 VPNs that use the IP protocol as the
network layer protocol. However, the deployment of these different VPN archi-
tectures is rather controversial because IPSec VPN is based on the overlay
model (see Section 1.4.1), while MPLS VPN is based on the peer-to-peer model
(see Section 1.4.3) which offers the combined benefits of traditional overlay and
peer-to-peer VPN models (see Table 1.2). The controversy between the IPSec
and MPLS VPNs is narrowed down to one important factor: who owns the VPN.
14 The VPN Overview
TABLE 1.1 Upside and Downside of Overlay VPN versus Peer-to-Peer VPN.
Overlay VPN Model Peer-to-Peer VPN Model
Upside Downside Upside Downside
Allows replicate Difficult to size Routing between All VPN routes
IP addressing intersite circuit sites is optimal are carried in the
Full isolation capacity Circuit capacity service provider
between Fully meshed sizing between IGP
customers circuit sites is not an Replicate IP
Secure VPN requirement for issue addressing is no
service optimal routing Simpler routing longer an option
Layer-3 CE configuration for Complex filters or
routing customers (no dedicated devices
adjacencies overlay mesh)
between sites
TABLE 1.2 Benefits of MPLS VPN
MPLS VPN Model
Combined benefits of overlay and peer-to-peer VPN models
Routing between sites is optimal PE routers hold only pertinent VPN routers
Allows replicate IP addressing Full isolation between customers
Secure VPN service No complex filters or dedicated routers
VPN1 6/9/03 6:00 PM Page 14

In other words, the implementation of these VPN architectures depends on
whether the VPN is managed by the customer or the service provider.
Customers who manage their own VPNs usually adopt the overlay approach.
Overlay VPNs are connection oriented. They are based on creating point-to-
point connections and not networks. In IPSec VPNs, IPSec tunnels are created
to provide point-to-point connectivity at the customer’s business site. This also
implies that IP connectivity is required to establish these point-to-point tunnels.
The best way to interconnect the sites is via the Internet through a service
provider who is responsible only for providing Internet connectivity between
the customer sites. In this case, the customers will have to build and secure
their own VPNs across the Internet by deploying IPSec. These site-to-site
enterprise-managed VPNs are also known as Internet VPNs. Because site-to-
site peering is required when implementing IPSec, scalability issues will arise
when the number of customer sites grows.
MPLS is typically offered as a site-to-site VPN service from a service provider.
The service provider builds and manages a private IP-based network and offers
multiple customers IP connectivity between their sites across this network. This
highly scalable connectionless architecture allows individual customers to view
the MPLS service as though they had an IP VPN connecting their sites. This
setup provides customers the same benefits of a layer-2 private network (see
Section 1.4.1.2), but with scalability and the easy management features of an
IP (layer-3) network, by eliminating the need for tunnels or VCs. Since MPLS
VPN runs across a private IP-based network rather than the Internet, the
service provider has the capabilities to offer differentiated levels of services and
service level agreements (SLAs) to its customers. A typical MPLS VPN deploy-
ment is capable of supporting tens of thousands of VPN groups over the same
network. However, because MPLS VPN is based on a service provider’s private
network, the reach of the service is constrained to the locations at which the
service provider operates.
1.7 Summary
There is a growing demand for VPNs. The different types of VPNs such as
remote access, site-to-site, and firewall-based VPNs have been examined. The
layer-1, layer-2 and layer-3 implementations of overlay VPNs have been
explained, along with the shared router and dedicated router approaches for
peer-to-peer VPNs. We also briefly examined how the MPLS VPN model oper-
ates and compared the benefits and drawbacks of overlay VPNs versus peer-to-
peer VPNs and listed the benefits of MPLS VPNs, which are the best of both
worlds. The last section of the chapter addressed the controversy on the deploy-
ment of IPSec and MPLS VPN architectures. The remainder of this book dis-
cusses implementing IPSec VPNs in Enterprises and deploying MPLS VPNs for
Service Providers.
VPN-in-Brief 15
VPN1 6/9/03 6:00 PM Page 15

Buildvpn1.pdf

More Related Content

Similar to Buildvpn1.pdf (20)

Recently uploaded (20)

Buildvpn1.pdf