Subject: Re: Severe I/O overhead?
To: Thor Lancelot Simon <firstname.lastname@example.org>
From: Manuel Bouyer <email@example.com>
Date: 07/18/2005 20:05:59
On Thu, Jul 14, 2005 at 06:26:37PM -0400, Thor Lancelot Simon wrote:
> Both Jed and I have noticed a huge increase in latency for small I/O operations
> in NetBSD/xen when compared to NetBSD/i386, even running in domain0.
> One easy way to see this is to time the extraction followed by the removal
> of pkgsrc with soft updates turned on. The rm -rf takes about three times as
> long in domain0 on both of our systems as it does running a "bare hardware"
> kernel on the same machines. The time isn't accounted as system time and I
> have to guess that it's going to interrupt handling.
> Does anyone see an obvious cause for this? A cure? I assume this is why
> even though sequential I/O is not *that* much slower, and CPU-intensive
> stuff obviously is just as fast, things like system and package builds are
> so much slower under Xen than under NetBSD/i386.
I could see two obvious cause: greather interrupt latency, or overhead in
the bus_dma implementation. Off memory I don't think the bus_dma implementation
is much different from the i386 version. Maybe a kernel profiling would
Manuel Bouyer <firstname.lastname@example.org>
NetBSD: 26 ans d'experience feront toujours la difference