Some Techniques to help bench-marking php applications to compare raw performance across servers, php versions and different run-times (php-fpm,swoole...)
When developing applications in any language, writing tests is only part of the equation; understanding how to measure performance can offer trans-formative insights into your code’s efficiency.
Benchmarking involves measuring the time taken by functions or operations to determine their speed under certain conditions. This practice helps identify bottlenecks and optimize them for better performance.
- Uncover slow-running code paths.
- Compare different implementation strategies.
- Verify the impact of changes over time.
k6 is a high-performing load testing tool, scriptable in JavaScript.
The k6 client command line allows you to generate many users directly from the client. This means that the k6 client acts as a controller (component orchestrating the test ) and a load generator (component sending the traffic to your application under test).
k6 can generate results like the below :
scenarios: (100.00%) 1 scenario, 500 max VUs, 1m0s max duration (incl. graceful stop):
* default: 500 looping VUs for 30s (gracefulStop: 30s)
running (0m30.5s), 000/500 VUs, 15000 complete and 0 interrupted iterations
default ✓ [======================================] 500 VUs 30s
data_received..................: 1.2 GB 39 MB/s
data_sent......................: 1.2 MB 39 kB/s
http_req_blocked...............: avg=492.12µs min=2.29µs med=5.64µs max=50.06ms p(90)=8.7µs p(95)=17.95µs
http_req_connecting............: avg=452.39µs min=0s med=0s max=49.96ms p(90)=0s p(95)=0s
http_req_duration..............: avg=10.07ms min=921.3µs med=2.93ms max=138.15ms p(90)=35.3ms p(95)=49.24ms
{ expected_response:true }...: avg=10.07ms min=921.3µs med=2.93ms max=138.15ms p(90)=35.3ms p(95)=49.24ms
http_req_failed................: 0.00% ✓ 0 ✗ 15000
http_req_receiving.............: avg=532.76µs min=131.8µs med=495.17µs max=19.35ms p(90)=741.11µs p(95)=818.23µs
http_req_sending...............: avg=113.89µs min=8.65µs med=22.06µs max=42.12ms p(90)=43.54µs p(95)=52.35µs
http_req_tls_handshaking.......: avg=0s min=0s med=0s max=0s p(90)=0s p(95)=0s
http_req_waiting...............: avg=9.42ms min=703.98µs med=2.32ms max=98.57ms p(90)=34.45ms p(95)=48.5ms
http_reqs......................: 15000 491.048934/s
iteration_duration.............: avg=1.01s min=1s med=1s max=1.15s p(90)=1.03s p(95)=1.05s
iterations.....................: 15000 491.048934/s
vus............................: 500 min=500 max=500
vus_max........................: 500 min=500 max=500
You can read more about this awesome tool by reading the official documentations
PHPBench is an open-source benchmark runner for PHP analogous to PHPUnit but for performance rather than correctness.
You can simply write your test classes , like you do with unit testing and run the tests like below:
vendor/bin/phpbench run tests/Benchmark --report=aggregateYou get something like :
PHPBench (1.2.15) running benchmarks...
with configuration file: laravel-10/phpbench.json
with PHP version 8.1.24, xdebug ✔, opcache ✔
\Tests\Benchmark\ComputeServiceBench
benchIterate............................I0 - Mo360.000μs (±0.00%)
Subjects: 1, Assertions: 0, Failures: 0, Errors: 0
+---------------------+--------------+-----+------+-----+----------+-----------+--------+
| benchmark | subject | set | revs | its | mem_peak | mode | rstdev |
+---------------------+--------------+-----+------+-----+----------+-----------+--------+
| ComputeServiceBench | benchIterate | | 1 | 1 | 6.743mb | 360.000μs | ±0.00% |
+---------------------+--------------+-----+------+-----+----------+-----------+--------+Check the github repository for this tool
You can use/write, simple, custom scripts that calculate execution time of common control flows and function of the PHP. Helps you to compare benchmark speeds (PHP execution times), performances of servers, web hosting and also across PHP versions.
this is taken from the example from the script from the repository 1 below
-------------------------------------------------------
| PHP BENCHMARK SCRIPT v.2.0 by @SergiX44 |
-------------------------------------------------------
PHP............................................. 8.2.10
Platform........................................ Darwin
Arch............................................. arm64
Server........................................ hostname
Max memory usage.................................. 512M
OPCache status................................. enabled
OPCache JIT.................................... enabled
PCRE JIT....................................... enabled
XDebug extension.............................. disabled
Difficulty multiplier............................... 1x
Started at..................... 06/12/2023 13:45:37.453
-------------------------------------------------------
math.......................................... 0.0935 s
loops......................................... 0.0121 s
ifelse........................................ 0.0173 s
switch........................................ 0.0172 s
string........................................ 0.1842 s
array......................................... 0.3212 s
regex......................................... 0.1769 s
is_{type}..................................... 0.0322 s
hash.......................................... 0.1202 s
json.......................................... 0.1586 s
-----------------Additional Benchmarks-----------------
io::file_read................................. 0.0129 s
io::file_write................................ 0.0715 s
io::file_zip.................................. 0.5335 s
io::file_unzip................................ 0.1571 s
rand::rand.................................... 0.0089 s
rand::mt_rand................................. 0.0089 s
rand::random_int.............................. 0.0679 s
rand::random_bytes............................ 0.2320 s
rand::openssl_random_pseudo_bytes............. 0.2953 s
-------------------------------------------------------
Total time.................................... 2.5214 s
Peak memory usage................................ 2 MiBwhile you can write your own custom script , i encourage you to take a look these two repositories for some benchmarking scripts example: