High Throughput Replication with Integrated Membership Management
summary
ChainPaxos is a chain-structured Paxos variant that aggregates accept and commit steps into a pipeline to cut per-node message cost and execute updates with constant message complexity. It provides local linearizable reads with no extra round trips by attaching pending reads to the next update, and embeds membership changes so no external coordinator is needed. Evaluation on Grid’5000 vs Zab/Zookeeper shows higher throughput and scalability, reads scale with replica count (≈1300→1600 RPS from 3 to 7 replicas) and latency stays low for small clusters.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:30Motivation and Context
- 2:40Performance of Read Operations
- 4:08Membership Management
- 6:56Chain Boxes
- 9:06Linearizable Reads
- 14:37Evaluation
related talks
16mSwiftPaxos: Fast Geo-Replicated State Machines
46mpg paxos Table Replication through Distributed Consensus
12mEPaxos Revisited
01h 14mThe hitchhiker's guide to distributed transactions
17mNear Optimal Latency Versus Cost Tradeoffs in Geo Distributed Storage
19mPineapple: Unifying Multi-Paxos and Atomic Shared Registers
16mBoson Consensus: A Scalable Blockchain Sharding Consensus
23mLightning Talks on Distributed Systems