tree 794f0776f57c3bd736376b284483881906f2c407
parent a31889026f7a4ddebdb03c2bdfbc83f5e07b0b1b
author Hunter Lee <hulee@linkedin.com> 1571449289 -0700
committer Jiajun Wang <1803880+jiajunwang@users.noreply.github.com> 1581107062 -0800

Increase parallelism for ZkBucketDataAccessor (#506)

* Increase parallelism for ZkBucketDataAccessor

This diff improves parallelism and throughput for ZkBucketDataAccessor. It implements the following ideas:
1. Optimistic Concurrency Control
2. Monotonically Increasing Version Number
3. Garbage Collection of Stale Metadata
4. Retrying Reads Upon Failure