I don't see the reason to get Jess involved at this point.

Mike

-----Original Message----- From: Chris Thompson [mailto:cet1@cus.cam.ac.uk] Sent: Friday, January 12, 2001 7:45 AM To: scl@sasha.acc.virginia.edu Cc: sirbruce@ix.netcom.com; jason.santos@onsemi.com; mphelps@cfa.harvard.edu; toasters@mathworks.com Subject: Re: Can volumes span FCAL controllers?

Steve Losen scl@sasha.acc.virginia.edu wrote:

[in response to Bruce Sterling Woodcock sirbruce@ix.netcom.com]

...

...

That's not entirely true -- you can span a volume across controllers, just dont span RAID groups across controllers.

When a volume has 2 RAID groups, is the NVRAM split among RAID groups? How are CPs done?

[ Good introduction to consistency points for newbies ]

So in answer to your question, the NVRAM is shared by all volumes and raid groups because it is a log of incoming write requests, not a disk buffer cache.

But to resume the original thread, the question is whether or not the writes done as part of CPs are clustered in a way that helps to reduce the overheads of switching between FCAL controllers.

CPs for different volumes are logically distinct operations, but are in practice synchronised, either by the 10-second clock or by NVRAM filling up. In saying that one should avoid spreading a volume over multiple controllers, but can have different volumes on different controllers, the assumption is that writes (mostly) occur first to one volume, then to another.

If the same is to apply to RAID groups, then the assumption is that the writes associated with taking a CP on a single volume are (mostly) clustered by RAID group. This sounds entirely reasonable, given that the filer tries to write whole stripes, or at least stripes in which as many planes as possible are being updated.

Chris Thompson University of Cambridge Computing Service, Email: cet1@ucs.cam.ac.uk New Museums Site, Cambridge CB2 3QG, Phone: +44 1223 334715 United Kingdom.