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.
Best approach of vps/dedicated comprehensive benchmark
What is most common benchmarking package (and command to execute with args) to test server performance including disk and network. See a lot of different benchmark outputs, but dont have any idea how to run same tests and compare results. And may be timed tests strategy?
Comments
http://bench.sh/ for a starter
I 'm very afraid to run scripts which I dont understand, and which are not from official repos at my dedi. Do any standard package exists?
https://github.com/K4Y5/ServerBench
yet another script you might not understand ;-)
but it doesn't do anything else then pull testfiles from different well-known providers and run dd three times to do a somewhat common benchmark.
there is no magic in that script, it is a simple sequence of commands you could put into your shell manually.
comparable in terms of ypu performance I'd suggest using geekbench v4 nowadays.
and if you really want to look further into disk-performance fio is most probably the way to go but much harder to understand than simple shell-script... ;-)
Try this one for even much more network locations
wget x86.ca/bench.sh -O - -o /dev/null|bash;-)
WHAT do you want to benchmark?
And I do not simply mean "disks" or "cpu" - I mean that a benchmark needs a context. Getting the numbers, for instance, from a graphics oriented benchmark doesn't tell me much for my, say, nginx plus postgresql server scenario.
My connectivity benchmark, for instance, for which you asked in the aruba thread, is great and fine for me because I chose the destinations that show the world like I'm interested in. The same benchmark would be all but worthless for a Chinese checking his HongKong server.
Plus: Keep in mind that you need - well designed and specified - series of benchmark runs (e.g. at different times of the day) unless all you want is to get first rough impression.
Is this actually benchmarking something or just the usual LowEndDickLength measuring?
If only there was more australian servers