Making the basic queuing model obsolete
roweber at IEEE.org
Fri Apr 7 15:30:31 PDT 2006
* From the T10 Reflector (t10 at t10.org), posted by:
* Ralph Weber <roweber at ieee.org>
My apologies in advance for taking Mark's original question
off on a tangent.
Would it be acceptable to your tape engineers to set the
various feature support bits the way SAM-4 and SPC-4 say
they need to be set for the Basic Queuing model, but
without having a magic name in the SAM-4 standard?
The moniker "Basic Queuing" was invented so that a SCSI
standard could say "Devices conforming to this standard
shall implement the Basic Queuing model" without having
to clone too much SAM text into the aforementioned SCSI
SAM-4 is clear about the setting of the Control mode page
QErr bits for Basic Queuing (mostly because those bits
existed when Basic Queuing was defined). If the corresponding
changeable bits are set to prohibit alterations in the QErr
bits, then the effect of Basic Queuing has been achieve
without naming it as such.
The other Basic Queuing features can be indicated using various
supported features bits which did not exist at the time that
Basic Queuing was added to SAM-2.
The Task Attribute support can be indicated in the SIMPSUP
and ORDSUP bits in the Extended INQUIRY Data VPD page.
Support for the Clear Task Set task management function can
be indicated in the response to the Report Supported Task
Management Functions command.
In short, everything that once made Basic Queuing special
has new, more fine-grained knobs to indicate its usage.
The newer indicators are better because hosts can test
what they care about directly and not depend on interpreting
statements in SAM. Since having two ways to indicate the
same thing is always bad, the less flexible feature, Basic
Queuing, should be removed.
In my opinion, the only possible justification for keeping
Basic Queuing would be if some SCSI Transport Protocol standard
was planning to need the implementation-restricting requirement
mentioned above. Frankly, I have some doubts that such a
situation would survive a T10 Letter Ballot.
All the best,
Banther, Michael wrote:
> Hi Mark,
> Some of HP's future tape drive products will use the basic task
> management model. It makes little sense to support the full task
> management model in some streaming devices.
> Michael Banther
> Hewlett-Packard Ltd.
> Telephone +44 (117) 312-9503
> *From:* owner-t10 at t10.org [mailto:owner-t10 at t10.org] *On Behalf Of
> *Evans, Mark
> *Sent:* 06 April 2006 21:29
> *To:* T10 Reflector
> *Subject:* Making the basic queuing model obsolete
> There has been discussion both in and out of T10 meetings for several
> meeting cycles now about making the basic queuing model obsolete. No
> one in these discussions has reported that they are aware of any
> devices that currently use this model. Does anyone know of any future
> plans for any devices to use this model? If I don't hear a
> resounding, "YES!" from someone in the next day or so, I'll add a work
> item to my list to create a proposal to make this model be obsolete.
> Obviously, if there are existing devices that use this model, those
> devices may continue to use it as described in today's standards.
> Mark Evans
> Maxtor Corporation
> 500 McCarthy Boulevard
> Milpitas, CA 95035
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org
More information about the T10