How to bring "Page Fault per Operation" value to normal state ?

[damode]

I couldnt find any mention about this particular topic anywhere, hence posting this question.

Currently, on the Search Head--->DMC--->Search--->KV Store:Instance, it shows "Page Fault per Operation" --> 226.85 and
on the Indexer it is, 162.44.

I am not sure if it actually an issue or normal because, there are no warning messages from Splunk instance, nothing came up related to this in Health check, no DMC alerts.

Please advise how can I bring this value to normal ?

[coltwanger]

I brought this down by adding more memory to my Search Head. I was seeing ~160 page faults per op and now i'm down to about 4 (32GB->64GB RAM). I agree there isn't a lot of information surrounding this topic. It appears the kv_store attempts to write to the page file on disk if it can't utilize memory, so theoretically either increasing your page file/swap size or RAM should help bring this value down.

[damode]

thanks for your response @coltwanger. I will try this.
Does this have a high performance impact on the system or highly alarming aspect about it ? Because the way it shows in the GUI hitting the value 226 which is way beyond its max value-2, looks quite scary.

[coltwanger]

Personally I haven't seen much of an impact whether it's 4 or 160. I just started exceeding my page file size which caused the KV Store to crash so that's why I started off in the memory department. I feel like that particular gauge is a little misleading with the values it's configured to report as "bad".