How to test two VPS/VDS providers?

Question

Network administration

Usually I put clients on a regular virtual hosting, recently one of them needed VDS, his project has a bunch of parsing processes.
I ordered hosting from provider A , posted the script there. I was a little surprised that the upload speed jumped from 2 to 3 times , I thought that an ordinary simple self-written script with a small number of requests to the database would not show such a difference. Was pleasantly surprised. But, the VDS itself was configured by myself according to the instructions on the Internet. Then I saw an advertisement for a provider that has a setup service and prices are about the same, but also a control panel to boot.
I asked those. support to set up VDS, posted the project there and ... was unpleasantly surprised that the performance was almost 2 times lower than on the VDS fromprovider A. Now we need to clarify that both hostings run on regular apache without any ngnix, hhvm, php-fpm and other stray things, on hosting B, which turned out to be weaker, they also installed OPCache, xDebug (does not start with each request)
Both VDS-ok 1 gig of RAM, 1 processor under the hood (I didn’t check which one, but on both tariff plans I chose a younger processor, and not the so-called powerful one)
As a result, the question is - guys, did anyone set the task of seriously testing VDS-ki? There is some kind of set of tools, such as Antutu for android, which you can run and test not only the physical system, but also the web server itself (how many requests per second it pulls, how quickly it establishes a connection with the database, how quickly it executes requests, to the disk as writes, etc., etc.)?.. I am sure that there are such load tools, I ask for advice from senior comrades.

Answer 1

And now the most interesting thing: How is the speed calculated?) Maybe just one of the servers is further away and it's just about delivering traffic?
And about VDS: does the client really need it? that is, does it have security requirements for hardware? if not, then VPS and go ahead (there will be no noticeable difference, but cheaper)

Answer 2

Apache Bench for example.
https://ruhighload.com/%D0%A2%D0%B5%D1%81%D1%82%D0...

Answer 3

I'm afraid that even with such tools, the tests will be very relative. It is not at all a fact that the N1 server will not experience peak load during yours, and the N2 server at that time will be placed on a newly commissioned machine and will simply “fly”.

but also a control panel to boot.

The panel usually consumes resources quite well, especially if it is a dead VPS with 1 virtual processor and 0.5-1 GB of memory.

which turned out to be weaker, also installed OPCache, xDebug (does not run on every request)

OPCache, if my memory serves me, has long been installed with PHP by default. But the installed XDebug on the production server... I'm even afraid to imagine what the author of such solutions uses.
I’ll tell you from personal experience, there was one such good X hoster, among other things, they allowed you to install an OS from your own image on the server, which personally made me extremely happy. But, one fine day, while raising another node for the needs of the project, I suddenly noticed that the system was installed for an indecently long time. I started to "pick" and noticed that the speed of writing to disk, on all my VPS's, averaged 5Kbytes / sec.!! 5 kilobytes! And then it generally fell to 1Kb on average, and then the entire DC apparently collapsed ... Fortunately, I managed to notice the problem in time and quickly moved the projects to another place.
All this I mean is that your one-time measurement will not give anything, except as indicators of something at the level of "now".

Answer 4

there was such a project https://twitter.com/ServerBear
the new one was not launched, there are some traces
https://www.lowendtalk.com/discussion/85392/buildi...