Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

global sort is experimental in v7.4 #15244

Merged
merged 1 commit into from
Nov 2, 2023
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions releases/release-7.4.0.md
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,7 @@ Quick access: [Quick start](https://docs.pingcap.com/tidb/v7.4/quick-start-with-
<tbody>
<tr>
<td rowspan="3">Reliability and Availability</td>
<td>Improve the performance and stability of <code>IMPORT INTO</code> and <code>ADD INDEX</code> operations via <a href="https://docs.pingcap.com/tidb/v7.4/tidb-global-sort" target="_blank">global sort</a></td>
<td>Improve the performance and stability of <code>IMPORT INTO</code> and <code>ADD INDEX</code> operations via <a href="https://docs.pingcap.com/tidb/v7.4/tidb-global-sort" target="_blank">global sort</a> (experimental)</td>
<td>Before v7.4.0, tasks such as <code>ADD INDEX</code> or <code>IMPORT INTO</code> using the <a href="https://docs.pingcap.com/tidb/v7.4/tidb-distributed-execution-framework" target="_blank">distributed execution framework</a> meant localized and partial sorting, which ultimately led to TiKV doing a lot of extra work to make up for the partial sorting. These jobs also required TiDB nodes to allocate local disk space for sorting, before loading to TiKV.<br/>With the introduction of the Global Sorting feature in v7.4.0, data is temporarily stored in external shared storage (S3 in this version) for global sorting before being loaded into TiKV. This eliminates the need for TiKV to consume extra resources and significantly improves the performance and stability of operations like <code>ADD INDEX</code> and <code>IMPORT INTO</code>.</td>
</tr>
<tr>
Expand Down Expand Up @@ -104,7 +104,7 @@ Quick access: [Quick start](https://docs.pingcap.com/tidb/v7.4/quick-start-with-

Before v7.4.0, TiFlash does not support window functions containing `PRECEDING` or `FOLLOWING`, and all window functions containing such frame definitions cannot be pushed down to TiFlash. Starting from v7.4.0, TiFlash supports frame definitions of all window functions. This feature is enabled automatically, and window functions containing frame definitions will be automatically pushed down to TiFlash for execution when the related requirements are met.

* Introduce cloud storage-based global sort capability to improve the performance and stability of `ADD INDEX` and `IMPORT INTO` tasks in parallel execution [#45719](https://github.com/pingcap/tidb/issues/45719) @[wjhuang2016](https://github.com/wjhuang2016)
* Introduce cloud storage-based global sort capability to improve the performance and stability of `ADD INDEX` and `IMPORT INTO` tasks in parallel execution (experimental) [#45719](https://github.com/pingcap/tidb/issues/45719) @[wjhuang2016](https://github.com/wjhuang2016)

Before v7.4.0, when executing tasks like `ADD INDEX` or `IMPORT INTO` in the distributed parallel execution framework, each TiDB node needs to allocate a significant amount of local disk space for sorting encoded index KV pairs and table data KV pairs. However, due to the lack of global sorting capability, there might be overlapping data between different TiDB nodes and within each individual node during the process. As a result, TiKV has to constantly perform compaction operations while importing these KV pairs into its storage engine, which impacts the performance and stability of `ADD INDEX` and `IMPORT INTO`.

Expand Down