That’s a real stupid way to do it, but thanks for the clarification
Not really that stupid, in the server world knowing how many cores a process or server is using is really the only CPU operation metric that makes sense because then you don’t need to know out of how many cores and all that. It also pairs well with metrics such as wait and idle to determine how a system is performing or where improvements can be made by using a single unit.
3 Likes
I disagree. As @Cmdrdpointed out its very useful information especially if you a running many servers.
2 Likes
Sounds like it will work just fine by default, as long as you set the CPU priority to low. If you still run into issues we could help you change your config to reduce the concurrent uploads, which will result in less CPU usage.
Please let me know how it turns out, and if there is anything we can do to help.
Are there some numbers on the s3 gateways throughput? It would be great to see some benchmarks around CPU requirements to sustain 50MB/s transfer speed.
Using s3 gateway with nextcloud I was only able to get a download speed of 25MB/s to my Desktop:
CPU usage was lower than with older version but still 20% of the system in short spikes or 1 core in spikes.
Looks similar to my benchmark. So 20% of 8 core system is around 160% CPU time usage. On average 25MB/s throughput needs 160% CPU time. I hope this number goes down in future 
There may be others tools that can accomplish the same task that may perform better. For example, we have an rsync tool that we are working on.
1 Like