For the last few weeks, we have using the current host bench marking method to test host and report it, which is in efficiency, we are now change to a newer way.
The previous bench mark test workflow is slow, which cannot quickly and automatically to generate a report – the speedtest.net multi-nation speed test has slowed down our test progress, because they only allow limited local region test server and specific not on the list server is not allowed.
On the other hand, Yet Another Bench Script has a better workflow in multi-nation internet speed test, they using iperf3 Network Speed Test hosts. But the only downside is they don’t have Australia Host and the closest Vetta.online, NewZealand host, their iPerf3 servers will only allow one iPerf connection at a time on each port (5200-5209). Multiple tests at the same time is not supported. And they are always busy (assuming you know what is busy in iperf3 means)
Now, the update#1 host bench marking report will be changed to following:
| Test Item | Test Agent |
|---|---|
| Disk i/o speed test | Fio(UNIX based OS) |
| CPU & task test | Geekbench (cli/w) |
| Local region internet speed test | Speedtest.net (cli/w) |
| Multi-Nation internet speed test | iperf3(cli), Speedtest.net (w) |
Due to speedtest-cli still limited to nearby server only, we will try our best to access web-gui to compete a Multi-Nation internet speed test, otherwise, setup private iperf servers in different regions will come into option.
| Test Region | ISP |
|---|---|
| Sydney, AU | Telstra |
| Auckland, NZ | Spark |
| Singapore, SG | Singtel |
| Hong Kong, HK | HGC |
| Tokyo, JP | GLBB |
| London, UK | Community Fibre |
| Los Angeles, US | Spectrum |
| New York, US | Spectrum |
This new work flow might help review topics to provide clear information and shorten the benchmark process.
1 thought on “Host Benchmark Scheme Update#1”