[Mashiro]

Hello Post4VPS Community

I am currently looking for a few people who are willing to spend a little of their time, a few GBs of bandwidth and a tiny bit of compute power to test the development branch of my bench.sh 2.0 server benchmarking script. With this branch I will be pushing fixes and new features before releasing them to the productive master branch. It is also simply meant to be a testing environment.

bench.sh 2.0 Server Benchmarking Script: https://github.com/hidden-refuge/bench-sh-2
Development Branch: https://github.com/hidden-refuge/bench-sh-2/tree/dev


Today afternoon I spent hours thinkering and working on a fix for a issue related to the disk speed tests and the calculation of the average speed of the disk. In all previous years and still even nowadays the disk speed has mostly never reached values of over 1 Gigabyte/s. Nowadays with SSD RAIDs and NVMe SSDs this is very much possible.

The script didn't account for that and so the calculation of mixed speed test results like let's "768 MB/s, 1.2 GB/s and 890 MB/s" was always messed up and incorrect. The GB/s values were not accounted for properly and the average speed was too low. This is what I was working on today.

My current problem is that I need testing of the new calculation routines on a real server or VPS that reaches over 1 GB/s on the disk speed test. I simply don't have such a server. I could only test the new calculation routines with hard coded values for speed test results.

How can you help me? That's actually easy. If you have a server that reaches such high disk speed please run a few cycles of the development version of bench.sh 2.0 on your server and post the results here.


How to:

Code: (Select All)

wget --no-check-certificate https://raw.githubusercontent.com/hidden-refuge/bench-sh-2/dev/bench.sh -O - -o /dev/null | bash

Upload the bench.log file to a pastebin service and post it here. Of course censor IP addresses, hostnames and other sensitive information first.

Like: https://pastebin.com/KE6Aauym

People are also welcome to make suggestions and report other issues.


Thank you in advance.

[deanhills]

My current problem is that I need testing of the new calculation routines on a real server or VPS that reaches over 1 GB/s on the disk speed test. I simply don't have such a server. I could only test the new calculation routines with hard coded values for speed test results.

Does VPS 9 qualify for the test @"Hidden Refuge"?

[Mashiro]

Well, I cannot really possibly know this without having tested the VPSs myself. @deanhills

I did check the reviews we have for VPS 9 and based on that information I would say that VPS 9 does NOT quality for this test because of this speeds:
VPS 9 Review - deanhills: 571 MB/s
VPS 9 Review - Lightdestory: 328 - 372 MB/s
VPS 9 Review - ikk157: 102 - 362 MB/s

Those speeds are all below 1 GB/s. So testing the development branch on them will make sure that the new calculation does work properly (if values still add normally and the average is correct) but it won't make it clear if the calculation properly does its job in regards of mixed speed calculation or calculation of pure GB/s values.

This said tests like this are still welcome. Just to make sure the new calculation method doesn't break the normal way of calculating speed below the GB/s speeds.


I just figured that people here might also have VPSs they pay for or from other sides. Maybe they have such speeds. I might as well find no one to test it as needed. Well, that's not entirely true as the person who reported the issue has tested it once already.

---

I added a feature that will display the average result in MB/s and in GB/s. This needs testing, too. To test this any server will actually work .

[deanhills]

My HostUS VPS doesn't have a disk speed of 1GB/s but I completed the test successfully anyway.  Specs of HostUS VPS says 1 GB/s, looks as though that is not the case:

https://pastebin.com/28DWZ9W8

Latency of the VPS is great however and speed awesome.  



This is the package I'm on (the most basic one with London location):
https://hostus.us/openvz-vps.html

[Sn1F3rt]

Well this is the Dev Benchmarking Script results for my VPS 9 Atlanta:
https://pastebin.com/eRzqvBJn

I was completely baffled at the results. Either I'm mad or blind or alright has to be one of them. Till the KVM upgrade, my I/O speeds were around 800Mbps but today, it's 100Mbps. I could hardly believe my eyes.

The speed too isn't as great as it used to be before the upgrade:

Code: (Select All)

Retrieving speedtest.net configuration...
Testing from ColoCrossing (107.172.25.164)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by ColoCrossing (Buffalo, NY) [0.04 km]: 33.115 ms
Testing download speed................................................................................
Download: 344.71 Mbit/s
Testing upload speed................................................................................................
Upload: 259.91 Mbit/s

Well I'd thought of reviewing the VPS after the upgrade, but it's sort of disappointing. Nevermind, I won't complain for this free resource and anyways it fits my purpose.

Regards,

[LightDestory]

Well this is the Dev Benchmarking Script results for my VPS 9 Atlanta:
https://pastebin.com/eRzqvBJn

I was completely baffled at the results. Either I'm mad or blind or alright has to be one of them. Till the KVM upgrade, my I/O speeds were around 800Mbps but today, it's 100Mbps. I could hardly believe my eyes.

The speed too isn't as great as it used to be before the upgrade:

Code: (Select All)

Retrieving speedtest.net configuration...
Testing from ColoCrossing (107.172.25.164)...
Retrieving speedtest.net server list...
Selecting best server based on ping...
Hosted by ColoCrossing (Buffalo, NY) [0.04 km]: 33.115 ms
Testing download speed................................................................................
Download: 344.71 Mbit/s
Testing upload speed................................................................................................
Upload: 259.91 Mbit/s

Well I'd thought of reviewing the VPS after the upgrade, but it's sort of disappointing. Nevermind, I won't complain for this free resource and anyways it fits my purpose.

Regards,

I too got a lower I/O after the migration from OpenVZ to KVM of my VPS 9.. But still we can't complain at all. The VPSes are working withput any issue and maybe they added a sorta of "limiter" for the migrated VPSes, they can't giveaway newer and dedicated  hardware to "OpenVZ ex-plans"

With KVM you can't ovcersell anymore.

[Mashiro]

Thank you everyone who has tested the new I/O calculation with GB/s support. I have added @deanhills and @sohamb03 as testers in the credits of the script (see source code at the bottom).

I will be working on new additional features and formatting of the whole script. More to test coming soon.

@staff I'd like to request that you don't close this thread. I will keep it updated whenever I need testing of new features and fixes. Thanks .

[tryp4vps]

I recently have got access to a VPS with a NVMe disk. So I think I may help here.

Here is the benchmark log:

Code: (Select All)

System Info
-----------
Processor    : AMD EPYC Processor (with IBPB)
CPU Cores    : 1 @ 1996 MHz
Memory    : 992 MB
Swap        : 0 MB
Uptime    : 11 days, 21:25,

OS    : Ubuntu 16.04 LTS
Arch    : x86_64 (64 Bit)
Kernel    : 4.4.0-21-generic
Hostname    : (hidden)


Speedtest (IPv4 only)
---------------------
Your public IPv4 is (hidden)

Location        Provider    Speed        Latency

Global
CDN            Cachefly    104MB/s    11.656 ms

United States
Atlanta, GA, US        Coloat        26.4MB/s     55.994 ms
Dallas, TX, US        Softlayer        32.2MB/s     44.167 ms
Seattle, WA, US    Softlayer        32.1MB/s     41.623 ms
San Jose, CA, US    Softlayer        51.8MB/s    24.815 ms
Washington, DC, US    Leaseweb         37.7MB/s     64.925 ms

Asia
Tokyo, Japan        Linode        19.9MB/s    124.604 ms
Singapore        Softlayer        7.29MB/s    199.055 ms
Taiwan            Hinet            7.97MB/s    180.313 ms

Europe
Rotterdam, Netherlands    id3.net        14.4MB/s        159.916 ms
Haarlem, Netherlands    Leaseweb        58.6MB/s        41.546 ms


Disk Speed
----------
I/O (1st run)    : 577 MB/s
I/O (2nd run)    : 1.2 GB/s
I/O (3rd run)    : 1.2 GB/s
Average I/O    : 992 MB/s or .99 GB/s

Let me know if you need more help for the disk speed calculation test.