This post is more than 5 years old
10 Posts
0
2325
October 20th, 2015 10:00
A few questions re: abandoned snapshots, VMware Backup Appliance, DDBoost and NDMP
Hi all,
I'm running NetWorker Server 8.1.1.7 on Windows Server 2008 R2 with a VADP Proxy server, vCenter 5.5, 71 guest servers, 3 ESXi hosts, 1 physical server, backup target to DD620. Primarily synthetic full backups.
I see snapshot files left behind fairly often. I can consolidate these easily enough in vCenter. Question is: Does NetWorker just abandon the process when it encounters an error and leave the snapshots behind for me to manually consolidate? A few times I was unable to consolidate (VMware 5.1 to 5.5 upgrade seems to have addressed this though - maybe the host reboots released a locked file).
The individual that set this up had it setup with a VADP Proxy, but I see this is a legacy option when setting up backups. Is there a benefit to this vs. using the VBA? What is/are the benefit(s) of using the VBA?
If I switch to VBA, should I do this before upgrade to 8.2 or after? I'm thinking after...
Is it worth it to setup the client distribution? I only have a few hosts with the client installed for Exchange and SQL backups and can manually install, but it sounds like it's easier to keep things up to date.
The reseller didn't sell the DDBoost that I understand should have been included. Some of our backups run into the day. How much will this help backup throughput? Will it deduplicate/compress before data is sent to the DD620? The one physical server is at a remote location that we backup over the VPN, however we only backup a single folder of data and I will have to rebuild the server from scratch if it fails. How would DDBost help us here?
NDMP - From what I understand, this is for backups from a NAS. We are using unified storage on a VNX. From what I understand, there are 4 "streams". If I kick off a restore, there aren't enough streams while the backup is running and the restore fails. And a checkpoint runs hourly. Thus any restore requires backups be suspended, correct?
Thanks,
David
Vakil
1 Rookie
•
35 Posts
0
October 28th, 2015 04:00
Recommendations:
Get DDBOOST, it will solve backup speed issues, if you want I can discuss this in deep dive.
I would recommend upgrading your NetWorker Server to NW 8.2.1.8 and not NW 8.2.2.
Deploy VBA 1.1.1.50 with thick eager zeroed provision (better I/O performance).
Also there are many things you need to understand like compatibility and other stuff, so I would recommend you upgrading using EMC professional services. If you need I can help.
What is your DD OS version.
I have not recently seen any of these orphaned snapshot issues maybe since a long time now.
What about your Exchange, which version, DAG or standalone, OS.
Many questions actually but best would be contacting EMC PS who can help you set these things.
dsain1
10 Posts
0
October 27th, 2015 14:00
Anyone?
dsain1
10 Posts
0
October 29th, 2015 08:00
Thank you so much Vakil.
I asked for DDBOOST and the purchase was denied. I'll try to push it through again. The benefits are so great for the small cost, just like the training request that was denied.
Why not 8.2.2? That was recommended by NetWorker support (once they finally contacted me - I have a Sev1 ticket that's open right now and almost 24 hours unassigned - and that's with 2 calls into them following creating and an email to the manager of customer service).
Why VBA 1.1.1.50? The current one was recommended...
I'm not really sure what the VBA will give me but I'm hoping for better backups and eliminating the orphaned snapshots. I have a downtime scheduled for Sunday morning and can fix the orphans at that time, but what causes them? The VDAP proxy?
I have an offer for assistance in upgrading (one of the support engineers said contact him directly when I'm ready and he'll help, but this is the same one that recommended 8.2.2).
dsain1
10 Posts
1
November 6th, 2015 12:00
Looks like the abandoned (orphaned) snapshot issue is known to support. I received a private release VDDK that still hasn't resolved it. Now working with EMC VADP and VMware SDK support. Would be nice if it were as simple as an update to something.
Have a colleague that was working on a restore and it took 3 calls to support - after 45 minutes it seems as if they need to drop the call - They must be paid per call instead of resolution. He finally figured it out had to use FQDN of vCenter server instead of IP. Silly.
After 2 weeks of working with and getting nowhere with EMC, I'm thinking of abandoning NetWorker and going back to V****. I loved V**** for it's ease of use. V**** doesn't leave orphaned snapshots that require me to shut my servers off (in a hospital! Gah), clone them (per VMware support - it works, but rename the original first then clone back to original name to a different LUN) and power up the clone. Then go back and delete the original plus SSH to the server and delete the remaining (orphaned) snapshots from the originating LUN. 14.5 hours during off hours plus umpteen hours during the work week. Fun!