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.
Comments
You are right to a degree. I do not advertise vpsbench. But then I wrote it to scratch an itch, I properly introduced it in a thread, and I made it open source. Keeping a link to it in my sig seems to be good enough for me.
You are also right to a degree re the disk test. I should have balanced them out better so as to offer a comparison between seq. and rnd. operations. Turns out that's not at all an easy thing to do and I kept it the way it is because I don't want VPSB to offer "post processed" results.
The values my benchmark produces do not say that random read/write is faster or slower than sequential. What they show is the performance of each test. If one compares one must not compare seq. vs. rnd. but machine A vs. machine B.
There is a technical reason which IMO is worth to do it the way I do it: OSs, drivers, controllers do quite a bit of "magic" to be (or seem) fast. That's why I employ real random for random reads and writes; The pattern of those is really unpredictable and the data that is written is also random, instead of e.g. /dev/zero or similar with pretty much all other benchmarks out there. That's why those often measure God knows what (e.g. the OS or controller caching) instead of getting the disk performance.
But I concede that those are technical deliberations and most people here don't care and prefer to have something that is easily comparable. That's another reason why I never tried to push my benchmark. I talked about it (and my reasons for it) and give it away for free, so anyone interested can have it. And that's it. Oh, and I actually made the effort to create something between a good man page and a small documentation. And I provided a (small) array of test results, e.g. from a weak AMD 350 as well as from a Ryzen 1700 and a few others (incl. spindle systems, NVMe, etc).
TL;DR My benchmark is not for everyone nor was it meant to become a standard. It's my take on properly benchmarking and I refuse to benchmark VPSs (or dedis) with what I consider nonsensical crap (like dd based scripts). Use it if you understand and like it - or don't if you don't like it.
Thanks for your interest
I remember some discussion about this - is the source code actually available for inspection now?
Yes, since months.
awesome, that is great to hear!
(how did I miss that?)
I can now say with some sincerity: thanks for your contributions here.
Welcome.
And: uhm, I said/wrote it in the thread quite early back then. Some people asked friendly but with mild pressure for the source code and within a few days I responded by open sourcing it and providing the link which leads to both, prebuild binaries (for convenience and because not everyone has the build environment) along with the source.
Btw, back then, rather than a thank you I got weird discussions e.g. why (how dare I?) the code isn't on github but on Yandex. All in all the experience considerably cooled down my interest to provide open source.
https://www.lowendtalk.com/discussion/155563/new-server-monitoring-tool-feedback-desired/p1
this is the thread you're talking about?
(maybe I missed something, I think I tuned out after a page of discussion about why you couldn't release the source code ...)
anyway, good stuff, thank you. FreePascal for the win!