Mention the impact of schema use on performance about benchmarks

[abenhamdine]

Perhaps it would be preferable to see if there's still a performance gap after fastify/fastify#192 has been merged

[delvedor]

It is all written in the docs :)

[abenhamdine]

It is all written in the docs :)

Yes you're right, but not so easy to find for a newcomer 😉
People often only read tl;dr or article on Medium, and don't go further.

[cagataycali]

Is issue has any purpose or advice, how we can improve this situation?

[abenhamdine]

I just have run my simple benchmark with fastify last release 0.27 (https://github.com/abenhamdine/frameworks-benchmark) and there's still a significant performance difference (roughly 15%) with or without schema validation :

$ npm run test-perf-fastify

> [email protected] test-perf-fastify C:\node-projects\frameworks-benchmark
> autocannon -c 100 -d 5 localhost:3000/big && autocannon -c 100 -d 5 localhost:3000/bigschema 

Running 5s test @ http://localhost:3000/big
100 connections

Stat         Avg    Stdev   Max
Latency (ms) 194.49 13.32   238
Req/Sec      507.8  26.2    533
Bytes/Sec    86 MB  4.59 MB 92.3 MB

3k requests in 5s, 434 MB read

Running 5s test @ http://localhost:3000/bigschema
100 connections

Stat         Avg     Stdev   Max
Latency (ms) 171.4   10.6    215
Req/Sec      580.21  39.6    600
Bytes/Sec    70.7 MB 5.45 MB 75.5 MB

3k requests in 5s, 345 MB read

So I still think it would be mentioned somewhere or someone could be disappointed with performance on big responses without enabling schema validation.