I've recently had the opportunity to do some side-by-side testing between Apache and Litespeed. Its for a very high traffic site, runs around 100 MBit traffic peak daily.
Site currently runs on a Centos 6 64 bit server running Centos's Apache and php56w packages from webtatic. Loads are consistently below 1 on server with the following specs
Intel Xeon 1270v3
8 GB RAM
SSD storage
When we divert site traffic to the new Litespeed server, which has the exact same specs as above, load runs around 4 with peaks up to 8. All CPU related, IO is fine.
Page load times ARE faster, no question, but the load has us concerned.
We custom compiled php 5.6. Settings are as follows:
https://cl.ly/0s1C1109442A
Is it normal for Litespeed to take up more CPU for the same code base on the same hardware as Apache? Is that the price we pay for faster page loads? We feel like we do not have much overhead with loads running at 4 with spikes up to 7.
Site currently runs on a Centos 6 64 bit server running Centos's Apache and php56w packages from webtatic. Loads are consistently below 1 on server with the following specs
Intel Xeon 1270v3
8 GB RAM
SSD storage
When we divert site traffic to the new Litespeed server, which has the exact same specs as above, load runs around 4 with peaks up to 8. All CPU related, IO is fine.
Page load times ARE faster, no question, but the load has us concerned.
We custom compiled php 5.6. Settings are as follows:
https://cl.ly/0s1C1109442A
Is it normal for Litespeed to take up more CPU for the same code base on the same hardware as Apache? Is that the price we pay for faster page loads? We feel like we do not have much overhead with loads running at 4 with spikes up to 7.