Google Summer of Code 2023

Implement CPU shares control for compute-intensive tasks

• Mentor name: Ruihang Xia
• Mentor email: [email protected]
• Project type: 350 hours project
• Difficulty: advance
• Skill needed:
  • Rust
• Description of project:
  • This project is about providing a task runtime that are able to control CPU shares available to each task based on the upper-level context, like task group or task id. This is a fundamental building block for GreptimeDB to implement multi-tenants resource control mechanism.
• Steps to evaluate an applicant for the project: TBD
• What the first 2 weeks of work would look like during the internship:
  • Get familiar with GreptimeDB's query engine and tenant system
  • Understand how a runtime works and begin to think about how it would be implemented.

Make resource management system dynamic configurable

• Mentor name: Ruihang Xia
• Mentor email: [email protected]
• Project type: 350 hours project
• Difficulty: intermediate
• Skill needed:
  • Rust
  • Networking
• Description of project:
  • The goal of this project is to implement a dynamic and convenient configure interface for GreptimeDB's resource management system.
• Steps to evaluate an applicant for the project: TBD
• What the first 2 weeks of work would look like during the internship:
  • Get familiar with GreptimeDB's existing configuration framework
  • Get familiar with GreptimeDB's public admin APIs

Support compound dictionary array for query engine

• Mentor name: Lei Huang
• Mentor email: [email protected]
• Project type: 350 hours project
• Difficulty: intermediate
• Skill needed:
  • Rust
  • Apache Arrow
• Description of project:
  • Extend our query engine to be able to process dictionary encoded data format. This is useful for reading and computing columns with a limit value set.
• Steps to evaluate an applicant for the project: TBD
• What the first 2 weeks of work would look like during the internship:
  • Get familiar with GreptimeDB's query engine
  • Get familiar with Apache Arrow

Add supports for Duration and Interval data type

• Mentor name: Lei Huang
• Mentor email: [email protected]
• Project type: 350 hours project
• Difficulty: intermediate
• Skill needed:
  • Rust
  • SQL
• Description of project:
  • Duration type measures a range of time in either seconds, milliseconds, micro seconds or nanoseconds. And Interval counts days between two time. They are powerful types to help modeling real-world time series scenarios.
• What the first 2 weeks of work would look like during the internship:
  • Get familiar with GreptimeDB's query engine
  • Get familiar with GreptimeDB's type system

Truncate Table

• Mentor name: Yingwen Yang
• Mentor email: [email protected]
• Project type: 350 hour project
• Difficulty: advance
• Description of project
  • Implement TRUNCATE TABLE for the database. TRUNCATE TABLE removes all rows from a table. It is similar to the DELETE statement with no WHERE clause but is faster than deleting by row.
• Skills needed
  • Rust
• Steps to evaluate an applicant for the project
  • Build and run the database in standalone mode
  • Add a new variant (Truncate) to the RegionMetaAction in storage/src/manifest/action.rs
  • Fix all compiler errors
  • Add a new unittest to the tests mod in storage/src/manifest/region.rs that writes the RegionMetaAction::Truncate action to the manifest and loads it from the manifest
  • Open a pull request against greptimedb:develop branch with a branch that you created from your forked repo

Fault Injection Test

• Mentor name: Yingwen Yang
• Mentor email: [email protected]
• Project type: 350 hour project
• Difficulty: intermediate
• Skills needed
  • Rust
• Description of project
  • Test the storage engine with fault injection.
  • Use fail-rs to inject fail points to the storage engine and trigger them in the unit test.
  • Implement a unit test framework to drive the fault injection tests.
  • Add more test cases
• Steps to evaluate an applicant for the project
  • Build and run the database in standalone mode
  • Use fail-rs to inject a fail point to the storage engine storage/src