UNSOLVED
Impossibly slow Exchange GLR Restore from DD
Hello,
We are currently backing up our Exchange 2010 (DAG) to Data Domain using EMC Networker and NMM 3.0. Backups run fine and are pretty fast. However, when trying a GLR restore, the restore process is impossibly slow. I can select the database in the NMM GUI and start the restore process. However, traffic between the DD and the Exchange Server is at 20 Mbit/s only! Knowing that the DB that needs to be mounted is 264Gb, it would mean that the mount would take 66 hours. Why is it generating all this traffic at a ridiculously slow speed? The LAN is Gbit (The DD has got 2 Gbit NICs teamed using LACP on a Cisco Switch).
I am new to Networker (it was setup by an external partner), but I have been looking everywhere for a possible solution.
Any hints in the right direction would be very much appreciated.
Kind Regards,
Stijn Mahieu
NetworkSupport-
61 Posts
1066
0
Posted September 30th, 2014 06:00
Was this ever figured out? im running into a similar issue. the browse of the restore takes 3 hours. the restore of the data takes 3 min.
NetworkSupport-
61 Posts
1066
0
Posted September 30th, 2014 07:00
Well thanks for getting back to me on it. i will try the RDB to see if that is any quicker.. sad as this is such a nice and streamlined part of the app!
im going to continue to bang my head against it and see if i can track anything else down.
DeaconZ28-2015
1 Rookie
•
252 Posts
1066
0
Posted December 15th, 2014 05:00
Me too. Also running into an issue where it keeps dismounting the GLR DB.
For me mounting a 1,5 TB database takes about 1 hour, than I can browse the maiboxes but between selecting a user and viewing his messages it can also take up to 15 minutes. After that like miken-rigyW said restore takes 3 minutes.
DeaconZ28-2015
1 Rookie
•
252 Posts
1066
0
Posted December 16th, 2014 04:00
Hi Jarekb, what hardware are you using? Are you restoring from Data Domain?
DeaconZ28-2015
1 Rookie
•
252 Posts
1066
0
Posted December 18th, 2014 10:00
Hmm... very similar, I'm running a DD670 with 4 Gigabit interfaces in a port channel. On the Exchange side, its Cisco UCS with Gigabit fiber uplink. All are on a dedicated secondary network. I even put the mount point and NWFS cache on a dedicated Flash LUN.
I'm not sure what the problem is.
smahieu
6 Posts
1066
0
Posted September 30th, 2014 06:00
Hello,
This has never actually been solved. I opened a ticket @EMC for this, and in the end, we dropped GLR in favour of RDB restores, which are considerably faster. They also provided me some guidelines for GLR and restores in general:
- The directory containing the EMC software needs to have at least 20% free disk space of the total amount of data you are restoring or mounting.
- Create more but smaller Exchange datastores, which speeds up the mounting process.
I don’t know if this may be helpful to you, but it did not really provide a solution to our GLR problem.
Kind Regards,
Stijn Mahieu