Kademlia - a Distributed Hash Table implementation to power the overlay network of BitTorrent



667 views BitTorrent Internals



Kademlia is a P2P Distributed Hash Table implementation that distributes the KV stores across nodes in a network and retrieves them without any central authority/database.

Representation

Every node in the Kademlia network is identified by a unique 20-byte SHA-1 hash. The hash function is also used in reducing the key to be distributed to 20 byte unique id.

Ownership

The node closest to the key is the one that owns the key and is responsible for holding it. But to define which is the closest one, we need to define a distance metric that could quantify the distance between two entities.

Requirements from a distance metric

  1. distance between the node to itself should be 0
  2. distance between two nodes should be a positive number
  3. d(a, b) + d(b, c) >= d(a, c)

any function that satisfies the above 3 can be used as a distance function.

XOR as a Distance Metric

Kademlia uses XOR as a distance metric and defines it as a simple XOR between the ID of the entities

d(a, b) = a XOR b

XOR can be used as a distance metric because

  1. a XOR a = 0
  2. a XOR b > 0
  3. d(a, b) + d(b, c) = a XOR b XOR b XOR c = a XOR c = d(a, c)

Visualizing distance

XOR is not a usual distance we can measure, and hence it requires a special way to visualize. XOR tends to turn like bits to 0, so when we XOR two 160 bits numbers, the like bits will turn to 0.

Hence, the more common the prefix between the two 160-bit IDs, the smaller the resultant value, hence the shorter the distance. Thus we can visualize the distance between nodes by placing them in a complete binary tree.

In this complete binary tree, the nodes having shorter distances will be placed closer to each other. Instead of creating a tree with a depth of 160, the tree is constructed only to the point it minimally disambiguates the entity.

Routing

Because there is no central entity, nodes must know how to route requests among themselves such that they always converge to the right node.

To ensure this, every node in the network keeps track of a few nodes in its routing table. These are not random, but very strategic.

Every node knows at least one node in each subtree that it is not part of. This means that the routing table may not have the address of the desired node, but it can lead us to one of the nodes present in its subtree.

By following a greedy approach, the nodes can route us, step-by-step, to the desired node. This is a classic Overlay network with its own routing.

Updating routes

When a node receives a message from any other node, it makes an entry in its routing table thus ensuring the table is auto-updated without any explicit intervention.


Arpit Bhayani

Arpit's Newsletter

CS newsletter for the curious engineers

❤️ by 17000+ readers

If you like what you read subscribe you can always subscribe to my newsletter and get the post delivered straight to your inbox. I write essays on various engineering topics and share it through my weekly newsletter.




Other essays that you might like


Exploiting and stealing from the BitTorrent network

406 views 17 likes 2022-08-19

Stealing is bad, but in a P2P network, it is a cakewalk. In this 7th video of the BitTorrent Internals series, we take ...

Kademlia - a Distributed Hash Table implementation to power the overlay network of BitTorrent

667 views 36 likes 2022-08-17

Kademlia is a Distributed Hash Table implementation and it is used as an overlay network for BitTorrent. Instead of talk...

The Piece Selection algorithm that makes BitTorrent fault tolerant

421 views 22 likes 2022-08-15

Performance of the BitTorrent network relies heavily on the order in which the pieces are requested by the peers. In th...

The Choke Algorithm that powers BitTorrent

517 views 23 likes 2022-08-12

One of the most important algorithm that powers BitTorrent is The Choke Algorithm In this 4th video of the BitTorrent I...


Be a better engineer

A set of courses designed to make you a better engineer and excel at your career; no-fluff, pure engineering.


System Design Masterclass

A masterclass that helps you become great at designing scalable, fault-tolerant, and highly available systems.

800+ learners

Details →

Designing Microservices

A free playlist to help you understand Microservices and their high-level patterns in depth.

17+ learners

Details →

GitHub Outage Dissections

A free playlist to help you learn core engineering from outages that happened at GitHub.

67+ learners

Details →

Hash Table Internals

A free playlist to help you understand the internal workings and construction of Hash Tables.

25+ learners

Details →

BitTorrent Internals

A free playlist to help you understand the algorithms and strategies that power P2P networks and BitTorrent.

42+ learners

Details →

Topics I talk about

Being a passionate engineer, I love to talk about a wide range of topics, but these are my personal favourites.




Arpit's Newsletter read by 17000+ engineers

🔥 Thrice a week, in your inbox, an essay about system design, distributed systems, microservices, programming languages internals, or a deep dive on some super-clever algorithm, or just a few tips on building highly scalable distributed systems.



  • v12.7.8
  • © Arpit Bhayani, 2022

Powered by this tech stack.