>So you are saying that the entirety of swap at
highest >priority was filled and you ran out of swap
or what, >exactly?

With different priorities, at SLES7, we found that
Linux would go into a loop and lock out all other
tasks after the first swap with highest priority
filled. This was at the RC6 level. It may have been
fixed in the RC7a level and we have not seen the
problem since SP1 was applied. Nor have we seen the
problem with SLES8.

We routinely run with 4 real 3390-3 swaps on RVA and
we have pushed Linux to 22 swaps. (Both are volume
test workloads). Since they are all on comparable
devices, we set them at the same priority and I/O is
balanced across all devices.

It seems to me with a hierarchical swap structure
(starting with the fastest as highest priority and
going down to the slowest) that each time you fill a
faster device then when your slower device kicks in,
the is a disportinate decrease in performance. I
suppose I would only use this scenario if I
infrequently used the lower priority, slower device,
using it as a safety valve, as it were.

If I had a consistent paging load because of memory
constraints, I would prefer that my swapping devices
were at the same priority (relatively high) so that
the I/O would be evenly distributed over the devices
and a page-in would get decent service.

(It would be nice if Linux changed the names to
"paging", their real function, rather than "swapping",
so the discussion doesn't have to drift between "swap
devices" and "paging load". But then again, if wishes
were fishes ... ;-) ).

=====
Jim Sibley
RHCT, Implementor of Linux on zSeries

"Computer are useless.They can only give answers." Pablo Picasso

__________________________________
Do you Yahoo!?
Yahoo! Finance: Get your refund fast by filing online.
http://taxes.yahoo.com/filing.html

Reply via email to