Framework Competition — Real RoadRunner

Real RoadRunner server, resident PHP worker: the framework boots once, then serves every request. End-to-end HTTP over loopback — includes the constant webserver overhead (see floor-rr in the dataset); framework differences smaller than that floor are below its noise. Single sequential client. Whether a row also carries part of the worker's boot depends on the pool's max_jobs, which the dataset records: the boot is amortised across the jobs a worker serves between recycles, and a pool that never recycles puts no boot into a row at all. The deployment model actually measured is stated with the table.

PHP 8.5.11 · Linux 6.8.0-139-generic · roadrunner mode · OPcache: yes · 1000 iterations per run over 10 runs · charts show the median as a faint bar + dot with the fastest → p95 range, lower is better

Boot model: the worker is never recycled (max_jobs=0), so no boot is added to these requests; the one-time recycle is measured in the startup chart

Framework startup — worker re-init (RoadRunner)
Framework startup — worker re-init (RoadRunner)
Total response times
Total response times
Resident worker memory
Resident worker memory

Feature benchmarks

Routing

GET / — dispatches a request through the router and returns a rendered template — no database access. The `/echo` band runs the same path with a constant handler instead, so what the template render costs can be read off directly rather than inferred.

Routing
ORM / Active Record

GET /items — loads one page of the 1,000 seeded rows through each framework's ORM / Active Record layer: 20 items plus a COUNT for the pagination total. The `-fast` band repeats that page with hydration switched off, so the cost of building one object per row can be read directly instead of inferred.

ORM / Active Record
Query Builder

GET /items-qb — builds the same page of 20 items with each framework's query builder instead of its ORM, so the two data-access styles can be compared directly.

Query Builder
REST API (JSON)

GET /api/items — serves the same page of items as a JSON response rather than HTML, which adds serialization to the ORM work. The `-fast` band repeats that response with hydration switched off, so serialization cost and hydration cost can be told apart.

REST API (JSON)
AOP (Aspect-Oriented)

GET /features/aop — runs a request through an interceptor pipeline — logging, retry and middleware aspects wrapped around the handler. Only frameworks with an AOP layer take part.

AOP (Aspect-Oriented)
Cache

GET /features/cache — reads a COUNT(*) over the 1,000 rows through the framework's cache with a 10-second TTL, so a hit costs no database work and a miss runs the query.

Cache
Database Events

GET /features/db-events — inserts one event row per request and lets the framework's database events fire around that write.

Database Events
Event Dispatcher

GET /features/events — dispatches an in-process event to registered listeners.

Event Dispatcher
Validation

GET /features/validation — validates a payload with the framework's own validator.

Validation
Config

GET /features/config — resolves a value from the framework's config repository.

Config
Request-Scoped Services

GET /features/request-scoped — resolves a service scoped to the request from the container.

Request-Scoped Services
Rate Limiter

GET /features/rate-limit — checks a cache-backed rate limiter.

Rate Limiter

Latency by endpoint (ms, trimmed mean — lower is better) — the worker is never recycled (max_jobs=0), so each number is the measured request with no boot in it

RequestWorkloadAzeraLaravelSymfonySpiralCodeIgniterCakePHP
GET /no DB — routing + template only0.1740.4780.2500.4750.6540.286
GET /echono DB — routing + dispatch only0.1580.3900.2150.4570.5330.227
GET /items20 of 1000 items (page 1, + COUNT)0.2971.090.8030.8551.090.801
GET /items-fast20 of 1000 items (page 1, + COUNT), hydrated rows not built0.2650.8040.7620.7871.040.692
GET /items/11 item by id0.2120.7290.4490.6430.9720.571
POST /items1 row upserted (sentinel #999999)0.2790.7640.6370.7251.110.713
GET /items-qb20 of 1000 items (page 1, + COUNT)0.2640.7810.4880.6630.9840.556
GET /items-qb/11 item by id0.2100.6370.3720.5960.8850.485
POST /items-qb1 row upserted (sentinel #999997)0.2520.7320.5460.6401.140.575
GET /api/items20 of 1000 items as JSON0.2400.8740.4910.7420.9440.483
GET /api/items-fast20 of 1000 items as JSON, rows not hydrated0.2180.6770.4260.6220.8790.467
GET /api/items/11 item by id as JSON0.1950.7530.3910.5940.8870.551
POST /api/items1 row upserted (sentinel #999998)0.2210.7490.5970.6810.9000.689
GET /features/aopno DB — interceptor pipeline0.3130.6760.4580.802——
GET /features/cacheCOUNT(*) of 1000 rows, cached 10s (miss = query)0.1670.4650.2290.5350.6920.304
GET /features/logno DB — buffered log handlers0.1680.4370.2260.529——
GET /features/retryno DB — retry policy0.1640.4560.2390.541——
GET /features/pipelineno DB — middleware pipeline0.1690.4330.2290.632——
GET /features/db-events1 event row INSERTed per request0.2270.7070.4750.8251.040.543
GET /features/eventsno DB — in-process listeners0.2120.5940.3070.6320.9680.426
GET /features/validationno DB — validator run0.1751.210.4160.5730.9760.391
GET /features/configno DB — config lookup0.1650.5360.2360.5110.7000.231
GET /features/request-scopedno DB — scoped service resolve0.1630.5320.2280.5330.6020.230
GET /features/rate-limitno DB — cache-backed limiter0.1660.5690.2370.5430.6080.243

Server floor — real RoadRunner over loopback: a bare resident worker that renders a fixed string (floor-rr) measures the IPC + server floor every row below also pays. Only differences larger than this floor are framework differences.