Frank,
I know more-or-less PPRC-XD (played with that a little bit).
I know, there is no warranty about "empty bitmap" of changes.
However only recently updated tracks should be out-of-sync, the rest
shouldn't be altered. Recently - I mean less than half-day. Remeber: two
batch session a day, no I/O activity on the volumes during time between
the sessions. So, *for sure* those closed, and half-day old dataset and
their VTOC/VVDS/BCS description are in-sync.
The real question is whether some possible inconsistencies in
VTOC/VVDS/BCS structures can influence on those "old" datasets integrity.
Regards
--
Radoslaw Skorupka
Lodz, Poland
Frank Krueger wrote:
Bruce , Radoslav
XD ist just asynchronous PPRC - and there is no guarantee when and what
will be copied .
It may take more then just a few seconds after the WRITE that the
data is copied - depending on the load and the link capacitiy.
The good thing is - you can switch easily from PPRC-XD to PPRC and vice
versa -
f.e. to have the data copied constantly , then goto SYNCH on a time with
less activity -
and then you may suspend and FLASH or take Backups from another system .
The switching process normally takes seconds only .
Basically it is used for migration or creation of some test data - not for
disaster recovery
cause the data is mostly almost fuzzy.
Frank Krueger
As I understand PPRC-XD, it can be one or a few I/Os behind the primary
volume. A dataset which was JUST closed on the primary (and thus valid)
may be exposed to problems on the secondary if the failure occurs before
all its I/Os (and the VTOC/VVDS update I/Os from CLOSE) can be
mirrored. But datasets closed more than a few seconds before the
failure should be OK (same 99%).
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
