Bruce Black wrote:
My current client has a V2X2 and is thinking about replacing it with a
Shark. SnapShot is used to snap 600 volumes in about 5-10 minutes.
The physical tape backups are done from the snaps and take about 8
hours. This DR process is fully tested and works great. My main
concern if we replace the V2X2 with the Shark is the DR process. Has
FlashCopy improved to the point that you can make a point in time
backup and physically move it to tape later? And can you FlashCopy
the entire box in a few minutes? If not, the DR process for this
client is going to get much more complicated. PPRC or XRC are not
options due to cost. Let me know your thoughts.
Tom, we have a lot of experience with all the various vendor's instant
replication functions, since we support them all in our FDR INSTANT
backup product.
In our experience, the elapsed time to do the SNAPSHOTs and the time to
establish the FLASHCOPYs is similar. In both cases it takes only a few
seconds per volume (your milage may vary <gr>).
The big difference is the architecture:
The STK disks use a virtual architecture, so that the SNAPSHOT is just a
matter of copying pointers in a table. The snapped copy takes up no more
space on the back-end disks, except for tracks that are updated. The
virtual disk data is compressed, so it takes up less back-end room than
the actual data would require.
The IBM and HDS disks copy the data in the background when a FLASHCOPY
is done. The ESTABLISH may be quick but the background copy may take a
while, especially if you FLASH many volumes. The FLASHCOPY architecture
makes the flashed copy look like the original disk immediately so you
don't need to wait for the copy to complete, however, your performance
may suffer if you try to do the DUMP before the background copy is
complete.
EMC SNAP actually supports both options. TIMEFINDER/CLONE does SNAP to
real volumes, similar to the IBM/HDS Flashcopy. TIMEFINDER/SNAP does
SNAP to virtual volumes. These are not really like the STK virtual
volumes, but they do only copy tracks which are updated on the original
disks. In our experience, TIMEFINDER/SNAP can be slow, but
TIMEFINDER/CLONE is better.
Just to complement: there's also TIMEFINDER/MIRROR. It is the best from
performance point of view. It works like internal PPRC, so the pair have
to be established in advance, and when synchronized can be splitted. It
takes approx. 0 seconds to get copy of your volumes. Your can then reuse
the pairs, the next synchronization process will take significantly less
time since bitmap of changed tracks is maintained on both ends.
Last but not least: consistency group are available for split.
AFAIK consistency groups for FlashCopy are constrained to single (and
whole) LSS.
--
Radoslaw Skorupka
Lodz, Poland
----------------------------------------------------------------------
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
