> I was discussing a number of things with Velocity software just
> yesterday and the recommendation given was to tune virtual
> storage down to
> the point swapping just begins and then set up 10 1 meg vdisk
> swap spaces
> rather than 1 10 meg. This provides a normal working set size
> and then will
> only allocate swap space that needs to be used.

It's more granular, given. Earlier versions of Linux would only move on
in the tiered list of swapping areas if the previous area was completely
full; I'm not certain that's still the case. It also would have the
advantage of not blocking in a highly critical section of code on the 1
outstanding I/O per device number issue that's part of the 390 I/O
architecture.

It would depend on how frequently you expect to swap, and your tolerance
for swapping in general. Multiple swap areas tend to do better in heavy
swapping, but have overhead involved in managing the areas. Amount of
space and the division of the spaces is another tuning area.

-- db

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to