[OpenAFS] Re: Windows cache rehashed...

Murawski, Rob rsm4@cssd.pitt.edu
Tue, 16 Dec 2003 12:44:44 -0500

No, The Windows client creates a cache file like the other clients.  The
cache file is opened as a memory-mapped file.  Thus the access to this
file works like virtual memory.  It is paged in and out of system RAM.

Depending on the algorithm used by the cache manager (and I have not
looked into the details) there may be excessive paging because areas
throughout the file need to be accessed which means they get paged into
RAM.  I believe Jeff's argument is that this algorithm may be poorly
designed for the cache sizes in use today and/or this method of access.
(It very well could be the same algorithm used on the Unix clients which
do not have memory-mapped files)

So the file *COULD* be in system RAM or *COULD* be on disk.  (And
*COULD* be both)


-----Original Message-----
From: Joe Buehler [mailto:jbuehler@hekimian.com]=20
Sent: Tuesday, December 16, 2003 8:58 AM
To: openafs-info@openafs.org
Subject: [OpenAFS] Re: Windows cache rehashed...

Jeffrey Altman wrote:

> The moral of the story is simply that the algorithms designed for the=20
> current cache implementation were appropriate for the scale of the=20
> problem at the time of their implementation.  They are no longer=20
> appropriate given the scale of the problem today.

I have found this thread on the Windows client cache very enlightening.

One question regarding something I am a little hazy on:

The Windows client caches data in physical memory, not virtual memory?
So if I set the cache size to X MB, I have essentially reserved X MB of
system RAM for the AFS cache, unavailable for other uses?
Joe Buehler

OpenAFS-info mailing list