Live data from Hacker News

Rendezvous Hashing Explained

randorithms.com

11–12 of 12 posts

Re: Rendezvous Hashing Explained

#11
post #6

> If our first choice for a server goes offline, we simply move the key to the second server in the list (which becomes our new first choice). It doesn't make sense: how do you move the keys from an offline server? The pictures in the post suggest that keys are not redundantly stored on other servers.

You can couple Rendezvous Hashing with a redundancy method like replication or erasure coding.

The default for Tahoe-LAFS (a distributed filesystem that uses a form of Rendezvous Hashing) uses erasure coding to split each file into (by default) 10 segments of which any 3 are necessary to reconstruct the file. Those 10 segments are then stored across the first 10 servers in the list. [https://tahoe-lafs.readthedocs.io/en/latest/architecture.htm...]

That way, even when servers go away with your data (whether due to crash or even network partitions(!)), you still have a decent chance to locate your data.

Post reply on HN