Eventual Consistency is Tricky

[John L]

There are several eventual consistency patterns in blockchain technology, with two of the biggest being:

1. Proof of Work – like Bitcoin. This approach is based on ‘probabilistic finality,’ where branches (forks) may appear briefly, but as new blocks are added, the longest chain is increasingly likely to be accepted as final.

2. Proof of Stake – like Ethereum 2 or Tendermint. Here, a supermajority of ‘validators,’ chosen based on the amount of currency they stake, reach consensus by voting on the true state, achieving deterministic finality once a block is confirmed.

There are also quite a few more experimental models. While they vary in design, they all really emphasize ‘eventual’ consistency, often requiring seconds or minutes to fully confirm the network’s state. Each approach tackles eventual consistency in a different way, really fascinating to study.

[Saurabh Dashora]

Cool insights!

[Daniel Moka]

Eventual consistency is one of the most challenging topics in distributed systems. Solid breakdown for the different types of it.

[Saurabh Dashora]

Thanks Daniel!

[Akos Komuves]

This was pretty cool 🤯 learned a lot!

Thanks for the mention!

[Saurabh Dashora]

Thanks Akos

[Raul Junco]

A distributed system can't exist without eventual consistency.

Great strategies, Saurabh!

Thanks for the shoutout, my friend.

[Saurabh Dashora]

Thanks Raul!

[Rajeev Sivaram]

Then there are systems like Google's Spanner that do consistency, at global scale. :)