[FLINK-30235] Comprehensive benchmarks on changelog checkpointing

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 1.17.0
• Component/s: Runtime / Checkpointing, Runtime / State Backends
• Labels: performance
• Assignee: Rui Xia
• Reporter: Rui Xia
• Created: 29/Nov/22 07:13
• Updated: 20/Nov/23 07:16
• Resolved: 20/Nov/23 07:16

Description

Changelog checkpointing is functionally usable right now. To make it as a productive feature, more comprehensive benchmarks are required. In this issue, I aim to answer the following two major concerns:

• The expansion of full checkpoint size caused by changelog persistence;
• The TPS regression caused by DTSL double-write;

By the way, I will also present other metrics related to checkpointing.

Issue Links

• mentioned in: Page Loading...