[
https://issues.apache.org/jira/browse/RATIS-2264?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17937626#comment-17937626
]
Tsz-wo Sze commented on RATIS-2264:
-----------------------------------
[~ivanandika], thanks a lot for filing this JIRA!
bq. Using RocksDB as the Raft log ...
We may use RocksDB as the the RaftStorage (including RaftLog and other
metafiles). For Ozone, it may be able to increase the performance.
> Exploration of Raft implementations variants
> --------------------------------------------
>
> Key: RATIS-2264
> URL: https://issues.apache.org/jira/browse/RATIS-2264
> Project: Ratis
> Issue Type: Wish
> Reporter: Ivan Andika
> Priority: Major
>
> This is a wish to discuss practicality and benefits of implementing Raft
> extensions that aims to increase performance.
> A data replication survey article
> ([https://transactional.blog/blog/2024-data-replication-design-spectrum#_hybrid_leaders])
> highlights other variants of Raft such as Linearizable Quorum Reads,
> Commutative Raft, Erasure Coded Raft.
> Another notable extensions are:
> * FlexiRaft: [https://www.cidrdb.org/cidr2023/papers/p83-yadav.pdf]
> * ParallelRaft: Used in PolarFS
> ** [https://www.vldb.org/pvldb/vol11/p1849-cao.pdf]
> ** [https://hengxin.github.io/papers/2021-JOS-PRaft-en.pdf]
> **
> https://muratbuffalo.blogspot.com/2025/03/parallelraft-out-of-order-executions-in.html
> * Using RocksDB as the Raft log
> ** Supported in some Raft implementations
> ** Previously discussed in https://issues.apache.org/jira/browse/RATIS-1327
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
