go bench read + write receipts/logs for parquet vs pebble

[jewei1997]

Summary

• Add benchmarks comparing pebble vs parquet receipt store read/write performance across realistic workloads (100 blocks with 10-100 receipts per block)
• using 3000 receipts/block and 5k blocks for writes and 2k blocks for reads (block range is ~100-500 blocks) though.

Writes

• both pebble and parquet do sync writes to WAL + async writes to underlying store
• receipt cache is not used in benchmark for more accurate comparison
• command go test -bench=BenchmarkReceiptWriteAsync -benchtime=1x -run=^$ ./sei-db/ledger_db/receipt/

Reads

• pebble iterates over each receipt (still uses bloom filters) while parquet issues a SQL-like filter query
• when hitting ledger cache both are fast as expected but outside of ledger cache, pebble super slow due to need to iterate over each receipt
• tested over varying concurrency, block range, diff filter criteria (topics, address)
• command go test -run='^$' -bench 'BenchmarkReceiptReadStore100' -benchtime=1x -count=1 -timeout=30m ./sei-db/ledger_db/receipt/ -v

Benchmark Results

Parquet faster for writes and reads 👍

Writes:
On detached pacific-1 RPC node (EC2 instance with EBS), getting 347,112.00 receipts/s writes using parquet, but only "only" 143,000 receipts/s using pebble

Reads:
Not even comparable. Parquet way faster. 20ms per log for pebble vs 0.05918 ms per log for parquet

[github-actions]

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	✅ passed	Feb 24, 2026, 9:50 AM

[codecov]

Codecov Report

❌ Patch coverage is 66.66667% with 38 lines in your changes missing coverage. Please review.
✅ Project coverage is 67.71%. Comparing base (1eb6794) to head (98a56be).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
evmrpc/ethbloom/bloom.go	75.36%	14 Missing and 3 partials ⚠️
sei-db/db_engine/pebbledb/mvcc/db.go	21.42%	10 Missing and 1 partial ⚠️
sei-db/ledger_db/parquet/store.go	0.00%	8 Missing ⚠️
evmrpc/filter.go	83.33%	0 Missing and 1 partial ⚠️
sei-db/ledger_db/receipt/receipt_store.go	92.30%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #2794       +/-   ##
===========================================
+ Coverage   58.14%   67.71%    +9.57%     
===========================================
  Files        2106       45     -2061     
  Lines      173289     6953   -166336     
===========================================
- Hits       100751     4708    -96043     
+ Misses      63596     1694    -61902     
+ Partials     8942      551     -8391     

Flag	Coverage Δ
sei-chain	67.65% <66.66%> (+9.54%)	⬆️
sei-db	68.49% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
evmrpc/bloom.go	100.00% <100.00%> (+6.38%)	⬆️
evmrpc/filter.go	70.36% <83.33%> (+1.29%)	⬆️
sei-db/ledger_db/receipt/receipt_store.go	71.89% <92.30%> (-5.08%)	⬇️
sei-db/ledger_db/parquet/store.go	64.68% <0.00%> (-1.76%)	⬇️
sei-db/db_engine/pebbledb/mvcc/db.go	62.77% <21.42%> (-1.21%)	⬇️
evmrpc/ethbloom/bloom.go	75.36% <75.36%> (ø)

... and 2062 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[jewei1997]

just a refactor, not new code added

[cody-littley]

LGTM. Take my approval with a grain of salt, since I don't yet have a solid grasp of receipt store architecture.

My main comment is to make sure we do long running tests as well (not a blocker for this PR though).