New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Load testing
Just wondering how best to load test to simulate 10k simultaneous visitors, given that this is quite a different metric than 10k simultaneous requests. Anyone got any tools and/or guidelines? I've tried loadimpact before, but it's pricey and I'm not sure how well it compares to a load from real users.
Comments
Acunetix maybe?
https://loader.io allows you to test with up to 10k clients for free as many times as you like.
Passler.com webstress software (free) can simulate 4K/Windows device/vps. So if you run it on 3 Windows Server VPS you could get up to 12k.
https://gatling.io/
https://www.redline13.com/blog/ uses amazon spot instances - aka it's super cheap :-)
In their amazing slider they even list some examples, like jmeter test for 10k users, and 100 m3.medium servers was a total of like $6-7
When using the tool, it tries to select the most optimal instance size and count depending on your settings.
Thanks folks. Some great solutions I had never heard of.
Remember lots of load testings are HTTP/1.1 only and don't support HTTP/2 HTTPS so if you're load testing HTTP/HTTPS keep in mind with HTTPS you will be testing HTTP/1.1 HTTPS and not HTTP/2 HTTPS. Lots of web servers are now moving to HTTP/2 HTTPS so testing probably should account for that. Only HTTP/2 supported load tester I know of is nghttp2's h2load https://nghttp2.org/documentation/h2load-howto.html
I use use h2load for HTTP/2 HTTPS load testing i.e. Caddy vs Nginx HTTP/2 HTTPS load testing https://community.centminmod.com/threads/caddy-http-2-server-benchmarks-part-2.12873/
For HTTP/1.1 load testing I use siege and my forked version of wrk at https://github.com/centminmod/wrk/tree/centminmod
Also check out locust.io need to setup yourself though - did some tests at http://wordpress7.centminmod.com/132/wordpress-super-cache-benchmark-locust-io-load-testing/ - with Vultr and Packet.net bare metal dedicated on hourly billing you could whip up a good cluster for locust.io
There use to be Blitz.io but they're closed down now i.e. http://wordpress7.centminmod.com/186/php-7-0-1-redis-caching-for-wordpress/
Just installed h2load , but I noticed something
when I curl it , or ab it , the transfer size is like 500 KB , but if I do h2load , it only reports as 12 KB traffic
my command was
h2load -n1 -c1 -m1 https://XXXX/
could you please give me any hint about it ?
Got specific example commands and output to compare ?
HTTP/2 HTTPS has HPACK header compression unlike HTTP/1.1 HTTPS so if you use non-HTTP/2 supported tools like ab or some curl versions you'd be connecting as HTTP/1.1 HTTPS not HTTP/2 HTTPS
Though ab and h2load transfers look about right to me for non gzip requests
ab = Total 43180 bytes of which 42460 bytes was HTML
h2load = traffic total 43044 bytes of which 462 bytes are headers and 42461 bytes is data