Enonic version: 7.13.3
OS: Linux Server
We have a basic cluster (cubernetes).
We’ve got a high consumption of RAM under high load of 1 million users load test.
How to avoid of that high consumption of RAM?
Enonic version: 7.13.3
Do you get OOM, or high GC? If not, there is no problem.
I’ll try to clarify the details.
During server render it takes a lot of RAM.
Render time without parts ~ (43ms - 100ms)
Render time with parts ~ (400ms - 1s)
Ram is being used for every page rendered, this is natural. It will automatically be freed by the VM’s Garbage Collector.
Page rendering speed is based on what you actually do on the page. You can speed up rendering by various caching techniques, i.e. using grid cache, or lib cache etc
Can I use {ssr:false} in parts to reduce server RAM usage on ssr render?
I’ve switched parts to {ssr:false}, now I’m going to push that release to high load test and see results )
Yes it works, with {ssr:false} in parts we do’t have crazy RAM usage on page render 
This is just offloafing rendering to the client, if everything still works, and you are happy with this approach - fine.