--- James Bottomley [EMAIL PROTECTED] wrote:
On Mon, 2006-10-16 at 20:40 -0400, Ric Wheeler wrote:
On February 12-13, we have put together a combined Linux file system
IO 2-day workshop in San Jose, CA. Note that the USENIX File System and
Storage Technologies conference follows us in the same venue, so we hope
to get some interaction between the two groups as well as leverage the
USENIX people to help us get this done.
For more information, please see:
http://www.usenix.org/events/lsf07/
Just to clarify, this event is a follow on to the Vancouver Storage
summit.
The only mentioning of Vancouver Storage summit in the web content
therein was, quoting from http://www.usenix.org/events/lsf07/cfp/ :
...
* Progress reports on implementation of features discussed at the
Vancouver Storage Summit
...
So by follow on you mean progress report only? Or will this
event also accept and/or discuss new material?
Although USENIX is helping us to run it, you don't have to be
a USENIX member to submit a position paper. The idea of the position
papers is to give the limited number of places (for storage we've got
about 20-25 and about the same again for fs) to people who have
interesting topics they need to discuss---so if you submit, be prepared
to make a presentation of it.
Can I make a presentation of this paper:
Serial Attached SCSI, An Architecture For Linux
-
This paper would start with an overview of SCSI (SCSI-3, that is), its
object oriented nature and why such is the direction of SCSI.
Then an introduction to SAS from this SCSI point of view will be given,
i.e. where it fits in the object oriented model, why and how. There may be
very little SAS technical introduction--a couple of sentences, something
anyone would understand and something sufficient for the latter sections of
the paper.
Then an introduction to SAS as an architecture in a SCSI stack would
follow. A layered, object oriented model will be presented, similar to the
one found in my code.
This will be accompanied with a SCSI architectural roadmap, the how and
why the architecture.
Then an overview of pure-SCSI drivers would be given (at this point the
paper talks about implementations at each layer of the storage software
stack). Those are implementations which hide the transport layer completely
in their firmware, and present a pure SCSI picture, a la SAM, to the
OS. How and why they do it and why it is better.
Then the paper would talk about what unifies those implementations, how it
can be done, and why it should be done this way. An introduction to SDI,
SCSI Driver Interface, would be presented.
There would be a section on a SCSI/ATA Translation, SAT, and a SAT
Layer (SATL). Where it fits, how and why. What its interface is and why.
The paper would include pictures and figures as necessary to show layers,
object oriented concepts and the like.
The thread is here:
http://marc.theaimsgroup.com/?t=11419758022r=1w=2
Luben