Evaluate the techniques described in Diva++: Dynamic Range Filtering over Hard Workloads to determine whether they would improve our storage engine. This includes reviewing the paper, building a prototype on a feature branch, and benchmarking it against the current implementation.
Goals
- Read and summarize the paper's key ideas, assumptions, and claimed results. Note how they map onto our current architecture and where they'd need adapting.
- Implement the approach on a separate branch, keeping changes isolated from master.
- Measure performance against the current engine at every layer of the stack, from low-level storage up through higher-level such as relational/SQL through TideSQL.
- Document findings, including performance gains or regressions, trade-offs, and a recommendation on whether to pursue a production implementation.
Out of scope
- Merging into master. This issue covers research and evaluation only; any production work will be tracked separately.
Evaluate the techniques described in Diva++: Dynamic Range Filtering over Hard Workloads to determine whether they would improve our storage engine. This includes reviewing the paper, building a prototype on a feature branch, and benchmarking it against the current implementation.
Goals
Out of scope