-
- Enthusiast
- Posts: 39
- Liked: never
- Joined: Dec 22, 2010 3:50 pm
- Full Name: MattG
- Location: Philadelphia, PA
- Contact:
Full backup performance
I have 3 files server VMs with this config:
- C: drive on VMDK
- 1TB E: drive on vRDM
- 1TB F: drive on vRDM
All 3 servers perform a weekly full without CBT (it is set to CBT), in Virtual Appliance mode with Hot-add, and daily incrementals. The incrementals work with CBT and are extremely fast.
The full backups are extremely slow. For example one of these hosts has 965GB of actual data and it took 45 hours to run at a rate of 12/MBs.
Are these normal numbers?
-MattG
- C: drive on VMDK
- 1TB E: drive on vRDM
- 1TB F: drive on vRDM
All 3 servers perform a weekly full without CBT (it is set to CBT), in Virtual Appliance mode with Hot-add, and daily incrementals. The incrementals work with CBT and are extremely fast.
The full backups are extremely slow. For example one of these hosts has 965GB of actual data and it took 45 hours to run at a rate of 12/MBs.
Are these normal numbers?
-MattG
Twitter: http://twitter.com/#!/matthewgraci
-
- Chief Product Officer
- Posts: 31749
- Liked: 7252 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: Full backup performance
I've split this into new topic not to derail the original thread you chose to post this into with environment-specific issue.
Yes, this might very well be "normal" for your backup server and source/target storage performance. But still 2x slower than in the worst lab I have access to and 50x slower then some customers having top hardware reported.
Also, don't look at CBT processing performance, this will always be extremely fast because of so little data being processed compared to full (there are typically only 1-2% of changed blocks daily in an average VM). Because of that, even on slow setups the processing rates will be awesome. Real performance testing can only be done on full backup.
Thanks.
Yes, this might very well be "normal" for your backup server and source/target storage performance. But still 2x slower than in the worst lab I have access to and 50x slower then some customers having top hardware reported.
Also, don't look at CBT processing performance, this will always be extremely fast because of so little data being processed compared to full (there are typically only 1-2% of changed blocks daily in an average VM). Because of that, even on slow setups the processing rates will be awesome. Real performance testing can only be done on full backup.
Thanks.
-
- Lurker
- Posts: 2
- Liked: never
- Joined: Apr 26, 2011 6:56 pm
- Contact:
Re: Full backup performance
MattG,
do you read/write from/to iSCSI when you backup?
do you read/write from/to iSCSI when you backup?
-
- Lurker
- Posts: 1
- Liked: never
- Joined: May 04, 2011 8:27 am
- Full Name: Joerg Stueger
- Contact:
Re: Full backup performance
Hello everybody!
I got a similar issue here.
We (full-)backup our vCenter 4.1 once a week on Friday. (22 VMs, ~4,5 TB).
Backup from EMC² AX4, SATA and SAS LUNs to a ML150 (singe CPU E5520, 12 GB RAM, Adaptec 6445, 4*2TB Stripeset)
First machine is backuped with 165 MB/s. (looks good I think )
BUT: Overall proacessing rate !!! 19 MB/s !!!
Any ideas ?!
Cheers
Joerg
I got a similar issue here.
We (full-)backup our vCenter 4.1 once a week on Friday. (22 VMs, ~4,5 TB).
Backup from EMC² AX4, SATA and SAS LUNs to a ML150 (singe CPU E5520, 12 GB RAM, Adaptec 6445, 4*2TB Stripeset)
First machine is backuped with 165 MB/s. (looks good I think )
BUT: Overall proacessing rate !!! 19 MB/s !!!
Any ideas ?!
Cheers
Joerg
-
- Chief Product Officer
- Posts: 31749
- Liked: 7252 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: Full backup performance
Hi Joerg, could be just some processing rate display glitch. Hard to say. Kindly please contact our support directly with any technical issues. Thanks.
Who is online
Users browsing this forum: AdsBot [Google], denispirvulescu, Google [Bot], joshima, Semrush [Bot] and 135 guests