-
- Influencer
- Posts: 13
- Liked: never
- Joined: Oct 21, 2010 1:47 pm
- Full Name: Kent Herr
- Contact:
v5 Instant Recovery
I have been messing with this new feature and my first impression is, great idea, but not really usable if your backups are located on a Data Domain DDR. Sure I can restore the vm, but when I try to run it, it never actually comes up. I am sure if I waited several hours, it would eventually come up. In addition, attempting to complete a restore from this same configuration, appears to be extremely slow. I attempted a restore a 20 GB vm and after 3 hours, it was 3% complete. This is off a DDR 690 attached to a Cisco UCS chassis running 10GB as well. Unfortunately the backup server is still running at 1GB for a few more days. Does anyone have any insight to these questions/issues? Thanks
Kent
Kent
-
- Chief Product Officer
- Posts: 31812
- Liked: 7302 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: v5 Instant Recovery
We did not test vPower with backups located on DataDomain DDR. We just started testing with a few other dedupe vendors though, and first results are promising. How good is performance if you try to put regular VM on DataDomain NFS share, and try to run it from there? Essentially I'd like to get an idea how DataDomain handles random read I/O from a large file.
-
- Chief Product Officer
- Posts: 31812
- Liked: 7302 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: v5 Instant Recovery
To give an idea about vPower NFS and Instant VM Recovery performance, and what it is capable of with backups located on "regular" backup storage, here are some results from one of our beta testers.
The machine boots in under a minute (I didn't get the exact time). Boot times seemed virtually identical with backups located on both the CIFS share and the Linux server. I did not time them, but I would say in both cases the system was ready in 40-50 seconds.
Here are the results of the 500MB file copy test (copying between two volumes of instantly recovered VM):
backup on Linux server -- 1min 40sec
backup on CIFS share -- 1min 40sec
Just for kicks I also ran the copy on the actual, live VM and the copy took 1min 45sec, so my Veeam instantly restored VM was actually faster than the original VM, although, to be fair, the Veeam backup server is running on a modern cluster with fast SAN and servers, while the original VM is on hardware that is ~5 years old and has a slow SAN with only a few spindles. Still, not too bad.
-
- VP, Product Management
- Posts: 6035
- Liked: 2860 times
- Joined: Jun 05, 2009 12:57 pm
- Full Name: Tom Sightler
- Contact:
Re: v5 Instant Recovery
I spent a LOT of time testing performance of V5 with the beta versions and I'm surprised at your issues. I really don't know much about the Data Domain DDR devices, how do you access the data? Via CIFS? Do you have current firmware on the DD?itherrkr wrote:I have been messing with this new feature and my first impression is, great idea, but not really usable if your backups are located on a Data Domain DDR. Sure I can restore the vm, but when I try to run it, it never actually comes up. I am sure if I waited several hours, it would eventually come up. In addition, attempting to complete a restore from this same configuration, appears to be extremely slow. I attempted a restore a 20 GB vm and after 3 hours, it was 3% complete. This is off a DDR 690 attached to a Cisco UCS chassis running 10GB as well. Unfortunately the backup server is still running at 1GB for a few more days. Does anyone have any insight to these questions/issues? Thanks
Obviously if it's taking 3 hours to restore 20GB then something is seriously impacting the transfer rate as you should be able to restore a 20GB VM in just a few minutes even with moderate performance. Our in-house restore rates are in the area of 30-50MB/sec from a CIFS share, when restoring to a VMFS volume, and 80-100MB/sec when restoring to NFS datastores.
It would be interesting to know how long it takes to copy the VBK file from the DDR to the Veeam server just via Explorer. If it takes hours, then you would need to figure out why.
-
- Influencer
- Posts: 13
- Liked: never
- Joined: Oct 21, 2010 1:47 pm
- Full Name: Kent Herr
- Contact:
Re: v5 Instant Recovery
I can access the data via CIFS or NFS. i have tried both ways and neither is worth using. I will definitely look into all the suggestions and report back. Thanks.I spent a LOT of time testing performance of V5 with the beta versions and I'm surprised at your issues. I really don't know much about the Data Domain DDR devices, how do you access the data? Via CIFS? Do you have current firmware on the DD?
Obviously if it's taking 3 hours to restore 20GB then something is seriously impacting the transfer rate as you should be able to restore a 20GB VM in just a few minutes even with moderate performance. Our in-house restore rates are in the area of 30-50MB/sec from a CIFS share, when restoring to a VMFS volume, and 80-100MB/sec when restoring to NFS datastores.
It would be interesting to know how long it takes to copy the VBK file from the DDR to the Veeam server just via Explorer. If it takes hours, then you would need to figure out why.
-
- Service Provider
- Posts: 5
- Liked: never
- Joined: May 29, 2009 3:41 pm
- Full Name: Resiliences
- Contact:
Instant Recovery Issue
[merged into existing discussion]
Hello,
I would like to know how much time take your Instant Recovery process to restore by example a windows 2008 R2 clean installed.
I experienced something weird on a lab. It takes almost 30 minutes to restored the VM.
"checking if vPower nfs datastore is mounted on the host" take 10 minutes and register the machines almost the same.
Veeam Support told that is typical for instant recovery but on a another client, that's very faster.
Regards.
Sylvain.
Hello,
I would like to know how much time take your Instant Recovery process to restore by example a windows 2008 R2 clean installed.
I experienced something weird on a lab. It takes almost 30 minutes to restored the VM.
"checking if vPower nfs datastore is mounted on the host" take 10 minutes and register the machines almost the same.
Veeam Support told that is typical for instant recovery but on a another client, that's very faster.
Regards.
Sylvain.
-
- Chief Product Officer
- Posts: 31812
- Liked: 7302 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: Instant Recovery Issue
Hello, it's definitely not typical. Very fast for me too, almost instant in my lab. May be there is something wrong with your lab. Thanks.
-
- Service Provider
- Posts: 5
- Liked: never
- Joined: May 29, 2009 3:41 pm
- Full Name: Resiliences
- Contact:
Re: v5 Instant Recovery
Hello,
I think that I have something wrong with the nfs veeam datastore mounted. When I try to browse it, it's very long.
The only thing special in my lab is that I dedicated one VMkernel on a dedicated vSwitch with a different subnet. My two ESX have this same conf and I put a cross cable between them.
Note that it also make failed my backup yesterday night on all VMs that are on the ESX on which I did the instant recovery...
Any Idea?
Sylvain.
I think that I have something wrong with the nfs veeam datastore mounted. When I try to browse it, it's very long.
The only thing special in my lab is that I dedicated one VMkernel on a dedicated vSwitch with a different subnet. My two ESX have this same conf and I put a cross cable between them.
Note that it also make failed my backup yesterday night on all VMs that are on the ESX on which I did the instant recovery...
Any Idea?
Sylvain.
-
- Chief Product Officer
- Posts: 31812
- Liked: 7302 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
- Contact:
Re: v5 Instant Recovery
I have removed logs you posted. For any technical issues requiring logs investigation you need to contact support directly with full log files, as explained when you click New Topic. Thanks.
-
- Service Provider
- Posts: 5
- Liked: never
- Joined: May 29, 2009 3:41 pm
- Full Name: Resiliences
- Contact:
Re: v5 Instant Recovery
Sorry, I haven't seen this info.
I have already sent this log to support.
Anyway, I restart Managment agent on the esxs and it looked, it fix it. Wait and see
Sylvain.
I have already sent this log to support.
Anyway, I restart Managment agent on the esxs and it looked, it fix it. Wait and see
Sylvain.
Who is online
Users browsing this forum: Bing [Bot], Egor Yakovlev, Google [Bot] and 115 guests