Performance

This page is the canonical benchmark reference for Allyourbase.

The measurements below are baseline numbers captured from local benchmark runs and should be treated as best-case guidance, not production guarantees.

Methodology

These API benchmarks come from internal/api/benchmark_test.go and run in-process with Go httptest against a local PostgreSQL instance.

• The benchmark harness uses setupBenchServer() to reset schema state, seed rows, and create the server.
• Requests are executed by benchRequest() with httptest.NewRequest, httptest.NewRecorder, and srv.Router().ServeHTTP.
• No real network sockets, TLS, load balancer, or cross-host latency are involved.
• Benchmark functions covered: BenchmarkHealthCheck, BenchmarkReadSingle, BenchmarkListDefault, BenchmarkListFiltered, BenchmarkListLargeTable, BenchmarkCreate, BenchmarkUpdate, BenchmarkDelete, BenchmarkBatchCreate.

Command used from the repository root:

go test ./internal/api -tags=integration -bench=. -benchmem -benchtime=3s

Hardware and Software Environment

Measured on 2026-02-17:

• CPU: Intel i7-9750H (6C/12T @ 2.60GHz)
• OS: macOS
• Database: PostgreSQL 15.14 (local)

Benchmark Tables

Binary Size

Build	Size
Default (go build)	36 MB
Stripped (-ldflags="-s -w")	25 MB

Includes the embedded admin dashboard, migration tools, and MCP server.

Startup Time

Mode	Time
External PostgreSQL	~310 ms

Measured from process start until the health endpoint returns HTTP 200.

Memory Usage (RSS)

State	RSS
After startup (idle)	20.5 MB
After 600 requests (health + schema)	20.5 MB

No memory growth was observed in this run.

API Benchmarks (httptest + local PostgreSQL)

Benchmark	ops/3s	ns/op	B/op	allocs/op
HealthCheck	704,338	4,392	7,792	42
ReadSingle (100 rows)	73,311	47,786	10,198	96
ListDefault (100 rows, 20/page)	30,848	122,002	36,333	608
ListFiltered (1K rows, filter+50/page)	16,765	213,918	77,380	1,419
ListLargeTable (10K rows, 100/page)	8,576	397,538	143,761	2,698
Create	7,776	533,087	14,096	163
Update	8,252	444,788	13,713	150
Delete	10,000	404,463	10,239	89
BatchCreate (50 ops)	382	15,687,150	272,462	5,665

Interpretation and Caveats

Throughput summary from this baseline:

• Health check: ~228K req/s (baseline, no DB)
• Single record read: ~21K req/s
• List (default): ~8.2K req/s
• Create: ~1.9K req/s (includes DB write + fsync)
• Batch create (50 items): ~64 req/s, roughly ~3.2K items/s

These numbers are useful for relative comparisons and regression detection. They are not SLAs and do not represent p99 latency, internet-facing throughput, or full production topology behavior.

For production planning, include network hops, TLS, proxies, connection pooling, dataset growth, query shape, and deployment-specific limits.