On 27 sept. 2013, you wrote:

I currently have a Power7 production system and a Power6 HA
system, running IBMi in single partitions.

The systems are configured for iASP mirroring using the Geographic
Mirroring feature of PowerHA for i Enterprise Edition.

Currently I suspend the Geographic Mirroring for a few hours a
night to perform a backup to tape on the Power6 HA system.

You still sometimes run a SYSBASE backup on your production system,
right ?

During this backup the Power7 production system remains available
to production users, but mirroring is suspended to the HA system.
Any changes to production data are queued up to be sent when
mirroring is resumed. After the backup completes on the Power6 HA
system, mirroring is resumed and mirror synchronization is
achieved in about an hour or so.

I want to come up with a way to eliminate the ¡°mirror
synchronization time¡± and still have a point in time BU to tape
of all our production data.

One idea I was thinking about was to come up with a way to perform
a FLASH COPY of the remote mirror (the iASP attached to the Power6
HA system), without having to suspend mirroring from the Power7
production system.

I do not know any FlashCopy feature with internal disks. Therefore, I
guess you are talking about external storage FlashCopy feature.

My question to you is, is this even possible to do this using the
PowerHA for i Enterprise Edition software on the IBMi? I imagine
it would require sufficient disk and another active partition on
the Power6 system¡Š Any thoughts on this idea or any other ideas
on how I could eliminate having to resynchronize the Geographic
Mirror as part of my nightly BU process????

As you say, you would require a third active IBM i partition member of
the cluster. This partition would host the target flashcopy IASP (so
with a third disk space, but disk storage Space Efficient features
can be used). PowerHA can manage flashcopy session. We run here a
similar configuration (but with external storage metro mirroring in
place of IBM i geographic mirroring, because we are full external
storage) for 12 production partitions.
The fact is that, to ensure all data is written to disk before the
flashcopy, the normal way to use flashcopy is to vary off the
production IASP before the flashcopy and vary it on immediately after
the flashcopy. If you cannot afford a production downtime, you can use
the Quiesce ASP functions (CHGASPACT command) to force writing IASP
data, suspending IO to production IASP disks, then run your flashcopy,
then resuming IO ro production IASP. Note that in this case, the IASP
vary on on third partition will run as it would do after an abnormal
system end.
Regarding backup to tape information, BRMS has some nice features which
help you, check out developerswork IBM site

This thread ...


Return to Archive home page | Return to MIDRANGE.COM home page