p2p testbed uses request tracing to demonstrate distributed techniques and algorithms using libp2p and request tracing
testbed.go has a main function. This function starts and sets up the tracing process and starts a server that displays the tracing information.
It then sets up this p2p network, it create new network with a set number of peers, and then connects those peers. In this case, our connectPeers function takes each peer and linearly dials/connects to all the peers that come after it. So peer 0 dials peers 1-19, and peer 15 dials peers 16-19.
Once that is finished, the main function calls "SendPing," this sends a message, called "PING", from the first peer (peer 0), to a random peer, let's say peer n. This random peer should return a message, called "PONG". When we send a message, what we are really doing is opening up what's called a stream to another peer. (Note to self, wtf is a stream). When we open this stream, we start a loop that listens for a response (handleStream). On the other side of the connection, when a peer 0 sends a request to start a stream with peer n, peer n recieves the message and since it is both a client and a server it knows what to do with it. It opens up a stream as well and creates a loop that listens for more incoming traffic. In the case of "SendPing", peer n recieves just one message from peer 0. It examines the message. It sees that peer 0 has send "PING", and in response it creates a message that says "PONG" and sends that message back to peer 0. Once peer 0, receives "PONG", it signals to the server to hang up on the stream. This hangup on peer 0 triggers a hangup on peer n. The stream is now closed.
'Chandy Lamport is a snapshot algorithm that is used in distributed systems for recording a consistent global state of an asynchronous system.'
Need to read more about the Chandy-Lamport implimentation on the p2p-testbed
p2p networking is probably best defined in contrast to the more popular ways of communicating on the internet. Communication online is typically client/server or request/response, that is, a client (like our browser on our personal computer) makes a request to a server (for example, we type http://www.google.com
into the browser, and it asks the google servers for the google.com webpage), and the server returns some information to the client (google sends back the data that makes up the google.com website, which the browser displays for us).
In a p2p network, any node can be both a client and a server. We can ask for data from other nodes on the network, and they can ask that data from us. There is no centralized source.
Okay folks, Kasey (aka @ramfox) here. I'm trying to learn how the libp2p package works, and how p2p works in general.
For those who don't know, Qri is built on top of the distributed web. That's right muggles, you didn't even realize it, but just by downloading the Qri electron app and opening the app or running qri connect
from the terminal, you are joining the rest of us wizards on the distributed web. Not only is Qri on the distributed web, but Qri specifically uses IPFS, aka the InterPlanitary File System. They have a specification and implimentation of libp2p, and imma learn how it works and what it does.
I started off checking out the libp2p specification: You should go take a read, this README is more or less a response to it:
It begins with a why do we need libp2p: they need a way for peers to communicate with each other over IPFS, but they also can't make assumtions about which networks/protocols those peers will use. Basically, let the devs choose how they want to party, and build a tool that is flexible enough to let different apps party the way they want: "In essence, a peer using libp2p should be able to communicate with another peer using a variety of different transports, including connection relay, and talk over different protocols, negotiated on demand." source
K, made it through section 1.1, one to 1.2... And that's where I got lost. Cause guess what! I don't remember anything from the networking class I took in college half a decade ago! Section 1.2 started talking about the goals for libp2p, and suddenly I was lost. They started listing all these different network protocols, like TCP, UDP, SCTP, RNG, TSM, SKT, and if you are nodding in understanding while reading those then you are just as lost as I am cause those last three aren't network protocols they are League of Legends teams.
So girlfriend's gonna go through all these network buzz words they are using so she can actually learn what's up. And if you are having trouble diving into this topic, maybe it will help you too :)
(Don't want to bog down my time rn with researching all of these right away, I'm pretty sure that most of it will come up organically as I read into the spec and implimentation)
Ethernet Wi-Fi Bluetooth USB
IPv4 - IP version 4 -> 32-bit address, gives us 4.2 billion ip addresses, made at a time when the internet was just a thing for academics IPv6 - IP version 6 -> 128-bit address, gives us 3.4×1038 addresses ip addresses Hidden addressing, like SDP
ARP - Address Resolution Protocol is a protocol for mapping an IP address to a physical machine address that is recognized in the local network (Learn Address Resolution Protocol) -> physical address known as a MAC address (Media Access Control) (What is a MAC address) DHCP -> Dynamic Host Configuration Protocol -> the protocol that allows you to request an ip address from a DHCP server, eg your computer requesting an ip address from your router, or your modem requesting an ip address from your ISP (What is DHCP?) DNS - Domain Name System -> Introduction to the Domain Name System Onion -> encrypted anonymous routing, used by the TOR browser (The Onion Routing) -> Onion Routing - Computerphile
RIP(1, 2) - older protocol - Routing Information Protocol used to route data packets by finding the best hop count OSPF - Open Shortest Path First - routing protocol for IP networks BGP PPP Tor I2P cjdns
TCP - Transmission Control Protocol - 'reliable and connection-based' -> needs to set up a connection a 'three-way handshake' UDP - User Datagram Protocol - 'lightweight and connectionless' -> unreliable, no error recovery or lost packet recovery, UDP and TCP: Comparison of Transport Protocols UDT -> UDP-Based Data Transfer Protocol - "is a high-performance data transfer protocol designed for transferring large volumetric datasets over high-speed wide area networks." wiki QUIC -> Quick UDP Internet Connection - website WebRTC data channel -> Web Real-Time Communication -> wiki
RMI - Javascript equivalent for remove procedure calls Remoting RPC - Remote procedure calls -> way for HTTP - >
What is NAT Traversal: NAT (Network address translator) traversal is the work around used to compensate the fact that there are more devices trying to use the internet then there are ip addresses for them in IPv4. Watch this video for a good explaination: How Network Address Translation Works, and may I recommend watching it at 1.25 speed. There are varing protocols for nat traversal
ICE -> Interactive Connectivity Establishment
According to the Macao Communications Museum website, which is randomly in english, although I guess that's not too weird: "Multiplexing is the process in which multiple Data Streams, coming from different Sources, are combined and Transmitted over a Single Data Channel or Data Stream."
So basically, a bunch of signals are trying to jam over one 'wire', multiplexing combines these multiple signals into one signal. This concept is not new, It's been around since the telegraph.
We use it because it reduces the resources needed to transmit multiple signals
In the TCP/IP and OSI models, multiplexing happens at the transport layer, and it lets multiple applications use one network connection simultaneously
Resourses to learn about networking, super usefulllll: video - Networking Crash Course article - The Internet, Networks, and TCP/IP
4.1.1 kad-routing -> P2P Networks - Basic Algorithms (watch at 1.5 speed)
Swarm Distributed Record Store Discovery Messaging Naming
What is a Distributed Hash Table? "A distributed hash table is a class of decentralized distributed system that provides a lookup service similar to a hash table: (key, value) pairs are storied in a DHT and any participating node can efficiently retrieve the value associated with a given key." DHTs characteristically emphasize the following properites:
- Autonomy and decentralization: the nodes collectively form the system without any central coordination
- Fault tolerance: the system should be reliable (in some sense) even with nodes continuously joining, leaving, and failing
- Scalability: the system should function efficiently even with thousands or millions of nodes
A key techinique used to achieve these goals is that any one node needs to coordinate with only a few other nodes in the system - most commonly O(log n) of the n participants
Overlay network: The set of links each node maintains (it's neighbors or routing table), together form the overlay network. The node picks its neighbors according to a certain structure, called the network's topology
ALL DHT TOPOLOGIES SHARE SOME VARIENT OF THE MOST ESSENTIAL PROPERTY: for any key k, each node either has a node ID that owns k or has a link to a node whose node ID is closer to k. We use a greedy algorithm to forward messages: at each step, forward the message to the neighbor whose ID is closest to k, when there is no such neighbor, we must have arrived at k. Sometimes called key-based routing.
Swarm is an sbstraction layer over connection management Had no f-ing idea what 'swarm' meant, what it was supposed to do. But then I read this issue: New and more accurate name for libp2p-swarm, and it's amazing how much less confused I am now looking at the swarm docs
The Peer Routing is the mechanism/logic for how we decide what peers to contact, in order to get to the destination peer. The Distributed Record Store, is the place where the list of peers that we can contact is kept (info includes their peerIDs and IP addressess) The Swarm (think 'switch' instead) is what opens/closes/manages connections and handles muxing. Once you have a swarm set up, you can open and close streams to other peers.
Yo, reading the docs on Circuit Relay, they are so good. Examples of multiaddrs and why they are useful
Distributed Record Store: Interplanetary Record Store spec
Okay, so after all that context, I'm finally going to look at the p2p-testbed and see if I'm less confused than before.