In this release we have increased the NGINX worker process count for larger nodes. We used to have a hard-coded value of 4 for all Hypernodes, which was based on early benchmarks from long ago when we didn’t offer large nodes like those in the Excellence line yet. But for nodes with more than a few cores and heavy resource utilization this was not really optimal.
The new worker process count will now be determined according to the following formula:
[vcpus, 4] | maxThis should improve performance and throughput a bit for large nodes serving a lot of I/O bound requests.