I’m currently working with a client looking to deploy Netbackup 7.1 for their 300+ virtual machines across 36 ESXi vSphere 5 hosts. The managed service provider has proposed a deduplication device based upon EMC Data Domain DD890. At the moment I’m looking at alternative options (DD is expensive!) –are there any experiences, white papers, case studies anyone is aware of with regard to native data dedup with NBU 7.1? Any other feedback would also be welcome!
An 890 is a HUGE box. Are you sure that is the right one? What is the total amount of data being backed up?
That said, Symantec's Dedupe is okay, and their sales people will call it that. "Good enough." What's nice is that you can use your own SAN storage if you need to, or get one of their appliances.
DD dedupes much better, and faster than a compareable unit from Symantec. They also have some other nice touches like CIFS and NFS options you can use for database dumps.
That said, it's a good idea to evaluate both products, but don't just rule one out over another simply on price. That is a wrong way to evaluate things. It's business requirements first, technical needs next; and determine which products meet the businessneeds, then look at price as a factor....
One thing you will get with Symantec dedupe soloution is client-side de-duplication where a DD solution need to transfer the entire amount of data before deduping.
In my experience in the field, the Data Domain solution does work well with NetBackup, but really if you use this combination you must incorporate Enterprise Disk and/or DD Boost to get the catalog information back to NetBackup at least, or the best case deployment would be to incorporate NetBackup Deduplication and use the DD box as the backup target; this way you can achieve deduplication in three locations (media server, client-side and on the target).
However, in the best case scenario above, if this method is adopted then a large degree of cost and support compexity can be removed by using NetBackup appliance/s; as Emiliano mentions above the 5020 appliance (32TB usable dedupe capacity) would be a like for like technology swap for the Data Domain technology. The NBU 5020 appliance allows 6x appliances to be pulled together as a maximum capacity (192TB total dedupe capacity); the major wins of using/positioning a Symantec appliance are: 1) on cost; DD charge for licensing for deduplication, replication and DD Boost on top of the tin, whereas the NBU 5020 replication and deduplication are covered by the NetBackup Deduplication Option licensing 2) Support complexity; the whole backup solution is owned by Symantec, so one support line and one "throat to choke" - IBM Global Services deliver the on site break-fix and the standard warranty delivers a 4hr response.
I have positioned a number of these up against DD and have sold NBU appliances more times than DD; reduced complexity in licensing and support, cost saving on licensing and still a very high performing solution delivered. Obviously only relevant to NetBackup customers though
I would not call distributed segment processing deduplication, BOOST moves some of the workload from the DD appliance to to the media server, doubling the injection rate of the DD appliance. Network load between media server and DD appliance is reduce.
I really can't find client-side deduplication mentioned in the BOOST documentation. Do you have newer documentation at hand ?
Not on hand at the moment. But it's a feature of Networker with DD and BOOST. So I guess it's EMC's answer to what NBU 6.5 and 7 does with the PDDO on a media server. But in the case of BOOST it also covers some application level "distributed segment processing," on the clients themselves.
Two things to note, it needs networker... Not sure how good that is, though I hear it's better than Legato. It's also limited to only certain apps and not all apps or platforms.
So ultimately it's a moot point since this is an NBU discussion ;-P
300+ virtual machine backup using NetBackup
As stated in above posts, you have NetBackup dedupe appliance as alternate option to DD.
In any option you will assess make sure that you have VTL functionality on the appliance. Because even though you appliance (either DD / Symantec) have great deduplication savings having data pushed to appliance over LAN is a big (legacy task).
I have positioned DD / NetBackup appliance in most cases and both works. DD is clear leader in that segment so they charge PREMIUM for that (everyone does). DD will be very costly and mind it DD890 which EMC has suggested is a MAMMOTH.
Also whatever option you go ahead with, make sure that you use VTL feature with inline dedupe.
Hope this helps.
OST over 10GbE is great. OST was designed to eliminate the whole tape/VTL argument and complexity.
But I still question why 890? I think someone is trying to oversell you!
Thanks for all of the feedback; I've checked the high level proposal and the total amount of data is around 70TB RAW, the proposal is actually for 2 DD860 (my mistage) one at each data centre, with 96TB useable storage.
The VTL option is not needed at all in a DD (We have several DD690's) and we are quite happy how they perform under Netbackup (With OST & Boost).
We have them connected over 10Gbe and I think you can't beat the overal performance...
But it indeed comes at a hefty price...