On Friday, 03/26/2010 at 12:45 EDT, RPN01 <[email protected]> wrote:
> Ah, but while that was true of a ?real? 3390, is that also now true of
emulated
> 3390?s which are split across varying numbers of essentially SCSI disks?
A
> single 3390 mod 27 might be split up over several 9 gig physical disks
in order
> to implement the emulation. Is the controller smart enough to be able to
start
> an I/O to each, even though the I/O?s were sent to the same 3390
address?
>
> The waters get muddier every day....
I'll agree with Mark that isn't very muddy at all. A "volume" has been an
abstract concept for quite a while now; controllers can do whatever they
want and it doesn't affect the rules for SSCH-style I/O. The controller's
goal is to give Device End back to the channel as quickly as possible once
the data is in a safe place (usually non-volatile cache) awaiting final
disposition.
PAV does not allow multiple I/Os to the same subchannel. Rather, it
creates additional "alias" subchannels ("exposures" in ancient parlance)
that map to the same volume. I/O to each subchannel follows the normal
I/O rules: you can't start another one until the current one finishes.
Very much like shared dasd on separate chpids, but done on a single chpid
instead. For guests that understand PAV, you can dedicate the base and
alias subchannels to the guest and it will discover and handle it.
Alan Altmark
z/VM Development
IBM Endicott
