> Would someone please shed light on the following for me. What is the
> dif from swaping to paging other than swaping is using the swap
> space? I can't seem to find much info on paging? I believe if I'm
> swapping heavily that's bad, but where does the pageing go and is it
> bad for perf? THANX!!!
> 
>  % Swap space used          0%

This tells you that you are not using any of your swap space, so your virtual 
machine is probably large enough that no Unix paging needs to occur. 

> Swap-in rate               0/s
> Swap-out rate              0/s

In traditional Unix systems, virtual memory is page oriented, but if the 
virtual memory system becomes sufficiently stressed, the kernel will attempt to 
move entire processes to swap to free up the maximum number of pages with one 
operation (it's a crude sort of block paging approach without actually doing 
I/O optimization). This number indicates the number of times this has occurred. 
In versions of Unix without virtual memory (2.9 BSD and that ilk), this was the 
only way to overcommit memory. Swapping on PDP11s was very common, especially 
on the smaller 11/05 and 11/23 systems. 

VM/HPO used to have this ability (separate swapping and paging, and the ability 
to prioritize one over the other to different disk areas), but that ability was 
lost with the transition to VM/XA (restored by a 3rd party set of CP mods). 
This transaction really impacted the sales of solid-state disk units -- 
swapping to SSD first was a really nice performance boost on memory-constrained 
systems when we still had a 16M cap. Even on the 1st generation XA systems 
(308x), HPO was a real win. 

> Page-in rate           0.066/s
> Page-out rate         50.133/s

This is normal dynamic paging activity, eg pages/sec. As someone else said, if 
this is database activity, it's probably sending things out and then rebuilding 
the page internally instead of trying to pull it back in from paging space. 

Reply via email to