Visualize Regions or Replicas

[breezewish]

Feature Request

Is your feature request related to a problem? Please describe:

We commonly face these issues:

• How many miss replicas are there? Do they distribute on a specific host?
• Is the leader distribution of a table balanced for each store?
• Is the read / write flow of a table balanced for each store?
• What's the proportion of the hot / cold data for each table / database?
• What is the distribution of replica sizes, or keys in replica? (so that one can decide a better region merge configuration)

Describe the feature you'd like:

Issues listed above can be all answered by the region info accessible via region command in pd-ctl. However pd-ctl only produce raw data. People usually need to rely on other tools (which uses region info data under the hood), like writing a script to analyze, or using SQL command to query distributions.

We can provide a visualization / analytics page in TiDB Dashboard for region info, so that it can be easy to know everything about region info. The detail functionality of the page is listed in this proposal: https://docs.google.com/document/d/1moQVhvIgqu_FWuv_UMB76AM5tETJkyeWnzI04wyevhk/edit#heading=h.8xipn1t7e2cb

The page can be implemented by using the awesome visualization library SandDance.

This feature is a subset of https://github.com/pingcap-incubator/tidb-dashboard/issues/548

Describe alternatives you've considered:

Teachability, Documentation, Adoption, Migration Strategy:

[IANTHEREAL]

Description：Visualize Regions or Replicas Category：Feature、Usability Value： Simplify the debug method of replica/region and provide more fine-grained observability of TiDB Value Score: 3

[baurine]

• The link about how to use pd-ctl
• The link about how PD works