[Feature] [BanyanDB] Support function in row merging

[wu-sheng]

trafficstars

Search before asking

• [X] I had searched in the issues and found no similar feature requirement.

Description

In the MVP of BanyanDB, all rows would be overrides when the latest data was inserted with the same ID. In advanced, it is better BanyanDB could provide a function based merging mechanism to merge previous and new data. Based on this, SkyWalking OAP L1->L2 distributed aggregation would not be necessary, and also would not require reading metrics from DB before flushing back, maintaining session.

Use case

This would highly advanced feature, which should not be considered in the first release. In longer-term, this would show much better performance as a native database.

Related issues

No response

Are you willing to submit a PR?

• [ ] Yes I am willing to submit a PR!

Code of Conduct

• [X] I agree to follow this project's Code of Conduct

[wu-sheng]

Considering this was added as a part of OSPP 2024, we need to consider. @hanahmily

1. Complete feature description
2. In-one-minute aggregation and downsampling workflow
3. Function (re)implementation solution
4. Testing methods for <3> and relative framework setup.

This feature should have larger scope than students can do in OSPP.