lkml.org 
[lkml]   [1999]   [May]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Adding more than 8 swap partitions
Mark Lord wrote:

> Peter Rival wrote:
> ...
> > Now I can add up to 15 swap partitions, not 32. I have 20 disks hanging off of an AS4000, with the possibility of somewhere in the area of 40 disks by the time I'm done. All of these disks are going to be configured the same - 256MB for swap, and the rest as a standard ext2 filesystem.
>
> Decide what you want to accomplish, and then look for a better way to do it.
>

I do agree that this is not the optimal way to do this, but I'm stuck with what I've got. I _need_ to keep the configuration the same.

>
> If you want fast swapping spread out over 20 disks, then configure
> the 20 swap partitions together into one or more RAID0's, and swap to
> a few BIG md-swap partitions.

I'm reconfiguring another system that we have that way, but unfortunately, it is an AS4100, so the PCI slots are much more limited than on the AS4000 (8 vs. 16). I will be more than interested to see what kind of performance data I get out of that configuration, especially as it relates
to the current configuration.

Seeing as how everyone seems to be trying to configure a better system, which normally is the right thing to do, I feel compelled to explain why not changing the configuration is so important. I am trying to benchmark ext2 performance against UFS and AdvFS through the AIM VII fserver
benchmark. We already have all of the Tru64 (UFS/AdvFS) numbers, now we just need Linux numbers. If I were to throw md at the system, it would definitely help the performance, but it wouldn't directly correlate to any configuration we have as yet tested.

Like I said, I do understand that the right thing to do is to md all the swap and ext2 partitions (or at least make several smaller ones). And I am going to do that, I just need to know if it is possible to get a true apples-to-apples comparison for this specific configuration under
Tru64. I will say that so far (on the shallow end of the pool, if you will ;) ext2 is more than holding its own. But I can't get to the high end of this comparison without being able to access all of these disks. I personally would be very excited to see Linux compare favorably, or maybe
even beat, a very high-end UNIX implementation on identical hardware.

Taking some previous advice, I'll be looking at /proc/mounts and /proc/swaps instead of df and swapon -s output. (Spending too much time running too many different systems that don't have that kind of access...) Sorry for being so long-winded...I'm just excited at all this, I guess...

- Pete


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:51    [from the cache]
©2003-2014 Jasper Spaans. Advertise on this site