tikv / tikv/pd

use telemetry to help improve schedule strategy

Open
#2,970 1 comment 0 reactions 0 assignees View on GitHub
status/discussion-wanted type/enhancement
Dominant language
Go
Stars
1.2k
Forks
783
Avg merge
5d 21h
Merged PRs (30d)
36

Description

## Feature Request

### Describe your feature request related problem
In TiDB 4.0, we have introduced the [telemetry](https://docs.pingcap.com/tidb/stable/telemetry#telemetry) to help understand how to improve the product. In PD, there are many schedule paraments. In the past, most of the time it was determined through some experience, and there was not a lot of data support. We can use telemetry to help us to improve the schedule strategy and explore the feasible direction, such as:
- How much cost can we save by cold and hot data separation?
- How much CPU consumed under the number of regions of different sizes?
- How long does it take for us to balance data to the new store?
- How appropriate should the default value of the store limit be?

and so on...

### Describe the feature you'd like
Through telemetry, collect some information that can help us solve actual problems without designing user information.

- [ ] supports telemetry in pd/tikv
- [ ] collect some informantion

Such as:
- [ ] Proportion of active regions
- [ ] Ratio of the empty region
- [ ] Distribution histogram of region size
- [ ] The snapshot size when balancing the region
- [ ] The capacity of TiKV and the used ratio
....

### Describe alternatives you've considered
Do not involve the user's private information,and have an option to disable telemetry

### Teachability, Documentation, Adoption, Migration Strategy
- Go, Rust

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.