docs: add DuckLake vs Iceberg benchmark report - #79
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 72f2a4db37
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| "flush_count": 100, | ||
| "final_rows": 90000, | ||
| "data_objects": 101, | ||
| "delete_objects": 406, |
There was a problem hiding this comment.
For every delete-10pct and mixed-append-update-delete sample, delete_objects is inflated to total_objects—for example, Iceberg COW reports 406 delete objects here despite using copy-on-write. The generating helper at test/integration/lakehouse_feature_benchmark_test.go:400 classifies any full key containing delete, but the sample prefix itself contains the scenario name, so every data and metadata object matches. These committed object-count results should be regenerated after classification is restricted to the filename or actual object type.
Useful? React with 👍 / 👎.
| Run the benchmark with the integration stack: | ||
|
|
||
| ```bash | ||
| docker compose -f test/integration/docker-compose.yml up -d --wait || true |
There was a problem hiding this comment.
When Compose fails to start or health-check MinIO, || true converts that failure into success; Bash's help true documents that it “Return[s] a successful result” and “Always succeeds.” The subsequent benchmark then calls skipIfMinIOUnavailable and exits successfully via t.Skip, so this documented sequence can appear to complete normally without running the benchmark or producing the requested JSON output.
Useful? React with 👍 / 👎.
1 participant
Closes #70.
Adds benchmark documentation and committed baseline result snapshots for the DuckLake vs Iceberg benchmark harness that was merged in #75.
Included:
docs/benchmarks/ducklake-vs-iceberg.mddocs/benchmarks/results/*.jsonThe benchmark harness itself already exists at:
test/integration/lakehouse_feature_benchmark_test.goValidation: