(Guest Post)
Since the announcement of Nimble Storage integration by both Nimble Storage (now a part of Hewlett-Packard Enterprise) and Veeam, I’ve been anxious to test out the functionality. The opportunity to perform all my Availability tasks (snapshotting, backup, replication, recovery) in a central management location, the Veeam Backup & Replication (VB&R) Console, is highly appealing to me. Not to mention the recovery tasks are more simplified. In the midst of my testing, which I’m still undertaking, I’ve encountered a few bugs and mishaps. As such, I wanted to share them with you as it may impact your environment.
My Environment
First, let me quickly share a little bit of the technical details of my environment:
Veeam B&R Server version 9.5 U2, on a Windows 2012 R2 VM within vSphere 6.0U3
I am testing on 2 Nimble CS500 arrays currently running NOS 3.7x
Bugs Encountered
While testing, there are a few bugs I’ve noticed, one with the potential of disrupting your VB&R backup and replication jobs.
1. First, upon adding a Nimble array to the VB&R Console, Veeam creates its own Initiator Group (IG) on the Nimble storage array you add to VB&R (IG name: veeam-FQDNHostnameOfAProxy-VeeamInitiator-0). This makes sense to be non-obtrusive to any IG already created & in use, but is a small surprise because it is not mentioned in any documentation I’ve read thus far (Veeam Nimble Integration Deployment Guide; HPE Primary Storage Systems Integration Guide). There is a NOTE in the ‘HPE… Guide’ (pg. 8) mentioning Veeam using its own initiator, but doesn’t go into details about it (creation, use, etc). Anytime a vendor makes any change to a customer system, the customer should be made aware by said vendor of such changes and the change(s) potential impact, if any. Again, makes sense Veeam creates its own IG, but a little surprised it isn’t better documented. That isn’t a bug, but certainly something I wanted to make you aware of.
2. I then created a new Volume on my Nimble array to test snapshotting through VB&R. For this new Volume to then get seen & added to the VB&R Console, a VB&R storage rescan needs performed (rt-click on the array in the Veeam Storage Infrastructure tab > storage rescan). Since my arrays are in “prod”, it takes quite a while to do a storage rescan (45-60mins) because of lots of snapshots I have. After the rescan, my new Volume still didn’t show in the VB&R Console. I’ll explain why it wouldn’t show a bit later.
3. The issue in #2 led me to find this one – after a couple storage scans in the VB&R Console and still no Volume showing, I thought I’d try removing the storage array from the VB&R Console & re-adding it in hopes my new Volume would then show in the VB&R Console. So, I removed the Nimble array from VB&R (rt-click > remove storage) and here is where I found another issue, and was a bit shocked – an Initiator Group I manually created some time ago, and actively use in my prod environment, was completely removed from my Nimble array by VB&R! Though this is a bug needing addressed, let me share why this may or may not be much of a concern for your environment. The issue seems to be related to IG naming convention. The VB&R-generated IG is named beginning with “Veeam-” (as I shared in item #1 above). The IG I manually-created had a name of “Veeam-Proxies“. So, the issue seems to be with naming -> if you have any IG named such that it begins with “Veeam-“, it will be deleted if you remove your storage array from VB&R. If you don’t have any IGs whose name begins with “Veeam-“, you should be fine, based off the testing I did. This may be of little concern as well because if you set up your storage to be a part of Veeam, how often, if ever, will you be removing your Nimble arrays from the VB&R Console? Also, based off my testing, your manually-created IG shouldn’t get deleted for simply having Veeam anywhere in its name. For example, I renamed my manually-created IG with the word veeam in it a couple different ways -> CO-Veeam-Proxies, VeeamProxies (no dash); and the IGs with those names didn’t get deleted when removing my storage array.
Let me now address the new Volume not showing issue from #2 above. The reason why my new Nimble Volume didn’t get added to the VB&R Console upon storage rescan was similar to the IG issue – naming. I created a Volume named veeam-test and it wouldn’t show. It seems the VB&R > Nimble Integration functionality just does not like when you have anything created on your Nimble whose name begins with “Veeam-“. Once I changed the name to something else, it the Nimble Volume then showed.
Conclusion
I’m really excited about the potential this integrated technology can bring to my organization. I like the idea of centrally managing both my Nimble storage and Veeam jobs in one location instead of having to go back and forth between the two. Also, for restoration processes, being able to use VB&R Console and tools (FLR, Instant VM Recovery, etc) is far better than having to go through a list of steps required to restore a whole datastore via Nimble. As I continue to test, I hope the technology proves more seamless and less ‘buggy’ when actually running my Availability jobs than what I’ve experienced thus far in the setup processes.
You can find & ping me on Twitter:
Shane Williford, @coolsport00
Sr. Systems Engineer
I ran into the same thing in v9 with the VNX integration. AFAIK, this isn’t a “bug” per se but a “known limitation” that operates just as you described.
Thanks for commenting Chip. Good to know it’s not isolated to HPE (in this case, Nimble) Storage Integration.
Thanks for commenting Chip. Good to know it’s not isolated to HPE (in this case, Nimble) Storage Integration.
Muy bueno el blog. Saludos
Thank you Guillermo!