How doesn't Hbase use any consensus algorithm like RAFT or Paxos?

Question

I know replication in HBase is done via append only log files, to maintain the same order of entries as that in master to get replicated to slave's WALs I assume there must be some consensus protocol. How is it designed so that there isn't any consensus protocol ?

By ordering for example, update a boolean column to true & then false. If updates are written to replica's wal files in reverse the final state could be true instead of false.

score 0 · Answer 1 · answered Aug 11 '22 at 23:49

0

Hbase uses a primary/secondary architecture where only the primary receives requests and thus can achieve consistency without a consensus protocol since writes all go to a single place.

Your question is similar to asking why does postgres not use consensus protocol? Postgres also has a primary/secondary style replication mechanism

answered Aug 11 '22 at 23:49

Asad Awadia

1,417
2
9
15

I agree that master has a consistent view always but different replicas can get updates in a different order if there is no consensus algorithm. – Yaswanth Tavva Aug 19 '22 at 19:05
No, they can't. Postgres can have multiple read replicas and they all get the updates in the same order and yet there is no 'consensus' algorithm – Asad Awadia Aug 19 '22 at 20:20
Postgres uses paxos actually. https://morioh.com/p/cf69b1b84f87 – Yaswanth Tavva Sep 06 '22 at 18:19
Postgres absolutely does not use paxos. That is an experimental extension that somebody is developing – Asad Awadia Sep 06 '22 at 19:19

How doesn't Hbase use any consensus algorithm like RAFT or Paxos?

1 Answers1