protonet primer
- 1. PROJECT “PROTONET” Primer for a new Internet On a new Zero-Configuration, Meshed, Ad-Hoc Wireless Network, Distributed Content and Anonymous Interaction Ali Jelveh Summer 2008 Project ProtoNet 1
- 2. PROJECT “PROTONET” Let’s #ackin’ rock! Ali Jelveh Summer 2008 The Idea: The idea behind ProtoNet is to create a next generation of networks, independent, anony- mous, local and basically uncontrollable both in size and content. It is -boldly said- the rebirth of the internet idea with means not available ten years ago. Just as the World Wide Web was designed to be a tool to connect scientist and allow them share information, ProtoNet follows the same basic idea, to allow a free flow of information from a user to any other user where a connection between these two is possible (or probable). As the grip of big companies, governments, and NGOs gets stronger on the inherently inse- cure and - by design - monolithic infrastructure of the web - and as technology (specially in the realms of wireless communications) evolves and becomes more efficient, cheaper and more widespread, an opportunity arises for so called information societies to once and for all break free from their dependence on carriers, governmental and non-governmental information mo- nopolies and to create their own - independent - information sharing networks using basic available means. The hope is that, with a return of a free flow of information - and the inherent gain of privacy, control and personal responsibility - more drastic changes in the way we do business, learn and share things will be possible. I personally believe that it can - and probably will be a catalyst for progress in all fields of human interest*. I hope you'll enjoy reading this document as much as I had fun writing it (not much really but someone had to). Any feedback, questions, critiques are very welcome. Proto: A combining form prefix signifying first, primary, primordial; as, primitive in form; protoplast, a primordial organism; prototype, protozoan. (ORIGIN #om the greek protos ‘first’) Protoplasm: The viscid and more or less granular material of vegetable and animal ce's, possessed of vital properties by which the processes of nutrition, secretion, and growth go forward; the so-ca'ed " physical basis of life;" the original ce' substance, cytoplasm, cytoblastema, bioplasm sarcode, etc. Project ProtoNet 2
- 3. The Architecture: As nature passed through its countless years of evolution it created excellent examples of what would be needed to design the ProtoNet architecture. From stem cells to basic nervous sys- tems, the Protonet idea is to - wherever needed - find natural designs and use their paradigms (expanded to the technical world) to solve our problems. While there are several similar ideas, our approach allows us to create a drastically different implementation. We don't need to spend resources on how to monetize this idea (though there are ways to do that shown in its relative section in this document), or on implementing a billing system. We don't need to spend resources on finding a way of controlling the net, as it is expected to be uncontrollable. We don't need to find a way to implement a darknet (a trusted peers only network) as it can be easily implemented on top of the ptn protocol by interested parties. Most importantly the goal is not to create a new distribution system for the existing internet (even though our first iteration will certainly be a hybrid system) as ProtoNet is supposed to replace (at least on locality) the existing infrastructure, which allows us to approach any up- coming challenges from a completely different perspective. But this part is about architecture, so here it is (it is a rough draft so bear with me): Before we continue let us have a quick look at the current wireless internet infrastructure: The standard wifi setup is basically just a way to wireless allow connectivity to the internet provided by your telephone or broadband provider. what it basically does is securely (or some- times not so securely - see WEP) relay your wireless stations internet requests to the internet provider and relay its answer back to the requesting wireless station. Project ProtoNet 3
- 4. your the internet provider WiFi Router ProtoNet Phase 1, the basic idea and the introduction of the hybrid system: The Idea behind protonet is easy: use the existing tightly weaved network of individual inter- net and network users and their respective wifi capabilities and utilize this infrastructure to create the next-generation internet. Sounds awesome, but what does that really mean? Well, let's look at big city neighborhoods: every building has now around 2-10 wifi networks created by residents to use their respective broadband access wirelessly. Now imagine what would happen if all those single access points would start to interconnect. We'd have a gigantic city wide network of interconnected wifi nodes, practically covering every corner of the city **. Now add some storage to every one of those nodes, and suddenly the network becomes effec- tively a gigantic file area network. What you would look at would then be both an infrastruc- ture for data messaging, data retrieval and data storage. Basically what the current internet is without the need for broadband access providers (as network traffic would be transferred - at best - at the wireless standard speed (currently 50Mbps, soon to be over 100Mbps and later in areas of 1Gbps)) or centralized hosting services.**** However, what can be described in a few lines in this introduction is quite a challenge in the real world. As the goal is to ensure a successful penetration into key markets the need arises to create an easy, low complexity migration scenario, preferably giving the user the feeling that Project ProtoNet 4
- 5. when buying or installling our product he keeps everything he has now and is rewarded with a few new, great features. In fact we don't want to just give him this feeling, we want the actual product to deliver exactly that. So to ensure a successful product we will start with a simpler, easy to implement solution that would - when we're ready - allow the system to grow and evolve into what it is supposed to be. The hybrid: The hybrid can be described in simple terms, it is a system that - without hassle - allows you to use the internet as you do now (surf, share, play etc.) but also automatically connects you to your neighboring nodes, has storage and understands the protonet protocol. It has just a few additional key technical features over the current router/ap implementation and would thus strategize the roll out as a disruptive technology, basically vertically - at first - extending and finally replacing the current technology. Project ProtoNet 5
- 6. your the internet provider PROTON Router other protonet units What it means is that while, when you're surfing, downloading or chatting you won't feel the slightest difference, it will have taken the first step, namely creating the infrastructure, to a whole new generation of applications and communications. All your request on current inter- net services will continue to run through your broadband provider, but the moment you surf to your ptn page (http://proton, or ptn://start or ... or actually use a program explicitly communi- cating over the ptn protocol) you will be communicating directly with your peers. You will see users connected to other nodes nearby, you can contact them, share files, have a videochat with them or just browse the protonet neighboorhod (which actually is just a mirror of your real world neighbourhood) for new content. But it doesn't stop here. One of the basic features of this ptn router/ap is the ability to allow and manage trusted and anonymous users out of the box. While most of the current wifi access points are encrypted or closed for anonymous use, protonet encourages not only sharing the ptn network but also your broadband connection within reasonable limits. And this sharing is rewarded: by sharing your ptn *and* broadband access you get access to other shared broad- band accesses for file downloads or uploads (this feature will be discussed later on). Both these features are enabled by default, a Quality of Service configuration makes sure that trusted users of the current node always get higher priorities on all connections (while a need for higher pri- ority on protonet communications might be debatable). Project ProtoNet 6
- 7. your the internet provider other protonet units PROTON Router trusted clients anonymous clients Content and applications on the ptn: As described above each node has data storage capabilities (starting at a single disk 250gb (as of today) configuration, to a RAID systems with higher storage capabilities and easily replace- able mediums), it also comes with a couple of applications: - the protonet dashboard displaying several information tidbits on the status of the network (currently connected clients, both trusted and anonymous, their transfer rates, also informa- tion on nodes and further levels of connected nodes). - a bonjour web-based chat client, allowing you to chat to all active clients in the neighbor- hood. - a configuration utility allowing you to allocate diskspace to trusted clients, set node informa- tion etc. - possibility for anonymous users to run a limited number of sanboxed web-applications (think of blogs, link lists or similar) Project ProtoNet 7
- 8. - ... How disk space is managed: Remember that part of the main idea is to create a large distributed storage facility, thus 80% of the installed storage is by default open to all users, every user can both upload and download files, a basic api allows other nodes to list files and their respective metadata. As the openness of the storage space would - without the necessary precautions - be highly susceptible to dis- turbing interferences, a few basic rules would need to be implemented, there will be a single file size limit (to ensure that single users don't take up all the space), files will have specific life spans (smaller files have longer life spans, larger files have smaller life spans, life spans depend also on popularity of files, popular files live longer, unpopular files will be kept until the space is needed and their respective lifespan has ended), and several other network health mechanism will be detected to counteract unhealthy usages (the health of the network is not dependent on content, but on usage pattern, flooding the network with unpopular files is one of those pat- terns). Even though the default configuration is hassle-free (easy) and allows everyone to drop files there, the possibility is given for any user to request secured disk space from the owner of the node. While the author assumes that most users will never change the default configuration, the node owner has full control of his disk space at any time, he can change the freely available disk space, he can give secured access (meaning no lifespan and size restrictions) and other privileges to any user he marks as trusted and also at any given time remove them again. He is essentially the ruler of his node if he wants it to be so. The idea of control and responsibility: While every node owner can superimpose his control on his node and its relevant functions, the basic modus operandi assumes that only a small fraction will take the time to control both the stored content and its users. This would in turn allow a anonymous content distribution system ... Also, with the possibility of control comes responsibility. Let's assume that at a given point in time, the storage space has been fully used and is not expiring anytime soon. If an user would then want to put content online he couldn't. Unless he assumes responsibilty and de- cides that his want is important enough to realize it: it would mean either adding storage space to the network by adding his own node, getting secured disk space from a node owner, facili- tating a node update through a node owner or finding another way to achieve his goal. The Idea is that the demand and supply of this ecosystem will dynamically balance itself. Project ProtoNet 8
- 9. Self healing: One of the basic functions of the protonet infrastructure will be its ability to heal itself, e.g. detect and counteract unhealthy nodes and behaviors. Excessive traffic will be punished, or better said, excessive and unhealthy nodes are being cut off from the system until they are healthy again, first for a short period allowing the owners to ensure a return to healthy usage and then for longer and longer periods. Unhealthy usage will be counteracted unpopular content however not, your last resort will always be to add your own node if no one is letting you use their disk space. As time moves on heterogeneous implementations of nodes will be highly encouraged as the idea of a heterogenic network makes the network much more hardened and unsusceptible to attacks and unhealthy usages... How to find content and node mutation: One of the main challenges is to find a distributed solution to searching and finding content, the protonet search mechanism relies on a couple node mutations, nodes become indexer nodes, or even higher lever super search nodes, indexer nodes are basically indexing other nearby nodes for contents (from ten to up to several hundred nodes) and indexed nodes know they have been indexed by whom, in absence of a super search node searches are being handled by those, they have a simple api interface. If you have special content not hosted by the stan- dard node api you will have the possibility to subscribe an indexing node to your content (how- ever only a very simple exchange protocol will be implemented - your host will need to deliver a simple xml file for indexing, reindexing periods can be set by you, for redundancy the indexer node will also make sure that at least another indexer node receives your xml for indexing, un- availability will lead to a deletion of your index after a period of time, you will however be able to readd yourself to the indexer node). Now for more thorough searches, nodes having the processing power and storage space will become super search nodes, they will be a second layer on top the indexer nodes, being able to search through the indexes of several indexer nodes at once, they basically mirror the indexing nodes contents. Project ProtoNet 9
- 10. skizze of nodes, indexer nodes and super search nodes. (awesome drawing skills of the author at display) Of course your dashboard always tells you what your node is currently doing and what good it is doing for the community and the protonet. On a meta level the idea is to mostly have a network of trust distribution of content, mostly blog based as it now for most illegal and legal downloads, meaning someone publishes an index on the best movies, and makes sure those are linked and available on the network. Or someone has an blog you like to subscribe to. How to ensure a fast and secure distribution of that content and node mutation: Another node mutation is the routing (routemap) node, routing nodes will essentially cache and store efficient routes through the network, for anything that is farther than two hops (tbd), they can also suggest or become highway nodes, basically a subnetwork of nodes allowing larger traffic to flow efficently from one part of the network to another. How to localize and node mutation (implementation and need uncertain): Another node mutation is the lighthouse node, it needs some outside administering but will allow for its owner to set its current longitude lattidude position, this will be used to check the current network topology against the real life mapping and verify and modify when needed the current topology and addressspace. As a side note: all nodes contain all the code for all different node mutations. Project ProtoNet 10
- 11. How to do maintenance on the network: This may be one of the most important parts of the system, its maintenance. There will not be (beyond the implementation and testing phase) a possibility for auto-updates. Firmware or software updates need to be done with direct hardware access, no over the net updates will be possible. No external ssh root access will be possible (unless of course you have your own node implementation). External access poses an immense problem to network safety and security as such they will not be permitted on standard user nodes. We might even implement a failsafe firmware restore that is completely redundant. Now the question arises: if no on the air updates are allowed how are they done? Well, pretty simple: through a network of trusted - let's call them super users, if have to find one near you you trust that updates your machine, or you do it yourself. We'll make sure that it's reason- ablely easy to find one on the protonet. So even the networks update mechanism works on a network of trust basis. The Doing of Stuff: Now that we have some common ideas about the protonet network and how it is supposed to work let's have a birds eye look at its first implementation. Working by example, the first nodes: The basic setup will be some mini-itx systems running some unix/linux flavor, first step will be a two nodes setup, with implementation of basic communications and APIs needed for basic protonet functionality, also trusted and anonymous usage tests, performance checks on the in- stalled QoS system and ensuring it keeps most of the router functionality intact. Next iteration is a three point setup with the first node mutations implemented (testing index- ing, supernodes etc.). When a reliable three point test system is in place there will be a 0.1 network rollout, the three access points will be placed in the immediate surrounding of the our current home location for testing, this will be the basic testing grounds for this operation. We will contact persons and homeowners we believe to be essential in their positions to ensure broad coverage. When the first three nodes are running basic tests are needed, file transfer speed, implement- ing the basic version of the dashboard, getting data on wireless coverage and coverage quality. Trying out the first node mutations, testing automatic on the air updates (as said above this Project ProtoNet 11
- 12. functionality will later be removed, it will only be used during the developement phases of the project). After having a running three point system and having successfully initiated the first node mutations we will add more nodes to the system, trying to cover a larger area. It will be essential to monitor the user interactions and wether they will adopt the dashboard strategy, maybe we will have to implement someway for them to easily remember the dashboard ad- dress. a couple of basic applications will the be distributed to the nodes, the file upload/ download functionality, and a basic bonjour like chat system. At this point we might setup a central user configuration and settings server whose main pur- pose will be to centrally store a users access data, his contacts and maybe his bookmarks, wether this should be kept like that forever is definitely debatable. Personally I would prefer that everybody carries his own profile with himself and thus will be able to access all his serv- ices, have his own bookmarks, contacts etc... We will the continually add nodes and continue the networks development, do our first hack attack scenarios (if that doesn't happen by itself) and continuously monitor the usage of the system. The following steps will be dependent on our on site situation assessment, wether there will be a continuing developing in that direction or a rethinking of the system is needed. Beyond the first nodes: = Protonet Cluster HH-OTTENSEN Protonet Cluster HH-ALTONA Protonet Cluster HH-OTTENSEN Protonet Cluster HH-ALTONA Internet Internet Protonet Cluster Berlin Protonet Cluster Berlin Project ProtoNet 12
- 13. = Protonet Cluster HH-OTTENSEN Protonet Cluster HH-ALTONA = x hops Protonet Cluster HH-ALTONA Protonet Cluster HH-OTTENSEN Internet protonet Protonet Cluster HH-HORN Protonet Cluster HH-HORN Protonet Cluster Berlin Protonet Cluster Berlin How you can help: 1. Code, though there is a hardware platform prototype running (ubuntu hardy, on intel atom hardware) the protocol doesn’t work as of now. Languages will be C/C++ or similar for the low-level functionality and Ruby for glue code. Ruby on Rails or Merb (or an alternative small-footprint Ruby framework) along with JS will be used for the front-end needs. 2.Hardware / Money for Hardware, the idea is to get up a mesh of 10-15 running nodes when the basic stuff works (the hardware should be identical to the pro- totype) and then extend it to 500 nodes by March 2009 each node costs ap- proximately 200$ to build as of today. 3. Find some way to finance this operation, through risk capital, donations, selling of nodes for cash or ? Project ProtoNet 13