as far as i know NetApp dedupe is sweep based, not inline like DD.  DD is a dedicated backup platform, NetApp is a general purpose NAS platform (very good platform). I think your  management is comparing apples to oranges.

I'm skeptical that you'll get impartial advice about replacing an EMC technology with a non-EMC technology on the EMC Support Forum.

I agree with dynamox, for obvious reasons, but I also recommend getting in touch with a Data Protection SE in your area to get more information regarding the DD enhancements. I can help you with that.

Thanks,

-Ed

I love that you posted that here

I'll be honest, I'm unlikely to be impartial on this but I was curious, so I took a look at this array.

What strikes me about it is that on their own website they don't even mention de-duplication as it's capabilities, I haven't dug any further (I'm sure it does talk about it somewhere) but that would suggest it's not something they want to shout about for whatever reason.

If it is post process de-duplication then I would steer well clear - I know that from bitter experience from other products.

Consider post process dedupe as a credit card and inline dedupe as a debit card.

You unwittingly run out of space when you backup more data than the speed that the system can post process dedupe at - it's really unpleasant (just like spending more on your credit card than you get paid).

Most of the systems I saw stored the raw backup AND the dedupe data in order to make restores faster, so actually stored more than a system without any dedupe at all - it was kind of ironic I thought

Maybe they made it work but you should ask (maybe they are doing inline - I don't know).

If you have a DD890, then a headswap upgrade to a faster/larger DD should be fairly straightforward and take a matter of hours to complete and continue with your backups, whilst re-using your existing DD storage shelf investment.

I guess that if Netapp are confident in it's abilities as a backup platform then comparing it (over time) in your backup environment on evaluation will tell you an awful lot about whether it is of equivalent stature for the backup/restore and storage utilisation you see now and ultimately whether the cost savings you are seeking are really there.

-- Evaluating "over a period of time" is key when talking about Netapp, it gives you a much clearer picture.

Before you decide the DD890 is maxed out at 85%, maybe a backup assessment from your local SE will help you stream line your existing investment in Data Domain technology.

Good luck with whichever route you take, Jonathan

Using a NetApp filer as a backup target to remove a data domain does not sound really good for me. Some things you have to consider:

• NetApp is a filer (CIFS / NFS)  any  is also used as a block storage (but VNX is better)
• NetApp is not using inline deduplication -> first all data is written to disks and afterwards if there is time and resources freed up it will be deduplicated.
• NetApp uses a fixed block-size of 8kb to do its deduplication - this will result in less effective deduplication factor than a data domain
• Replication between two NetApp filers are not as smooth as doing a replication between two Data Domains regarding the bandwith used
• TSM is a bit tricky doing deduplication - just read the latest white papers about TSM - there should be some optimizations available to boost tsm and data domain a bit
• The last and most important: NetApp is not EMC

I would advise you to stick to a Data Domain when needing a Backup Target. NetApp has some use-cases but EMC in total offers the better story

do some reading on TSM dedupe implementation. You will need a much more powerful server, device class has to be file.

Thanks a lot for valuable advice.

Here is the development about Deduplication. It sounds right that E series doesn't have such function. However, I was told that we can leverage TSM backup deduplication.

How would you compare backup software dedup with storage dedup?

jbrooksuk made a good description on inline-deduplication and post-deduplication, but both of them are on the storage level.

In the backup software deduplication, it should be viewed as inline-dedup? and should it be good as well? I know it would take some CPU resources, but a large server (assuming a media server here) should not be a problem, and  I was told the cpu time will be minimum.

some FAQs for ya Welcome to Wikis

It seems TSM can do the deduplication on the storage, so, the question is, which one is better between TSM software and Storage dedup. That probably really should be determined by experience.

Anybody have experienced these 2?

I ran TSM 6.3.3 but did not get a chance to try dedupe as we migrated to Avamar. What i can tell you is that my TSM server had a 700G database (as you know it's DB2) and during housekeeping jobs like migration, reclamation, DB2 was generating a lot of I/Ops.  TSM server was running on VMAX 20k with 300G FC backened.  Without dedupe TSM was working really really hard, beating the snot out of the spindles.  I can only imagine how much more load will be added if you wanted to do server side dedupe.  So instead of buying expensive primary storage that has to keep up with that workload, let the Data Domain box handle it ?   By the way did you see that you can also enable client side dedupe with TSM, that would be an interesting option to consider as well.  I would personally prefer that versus server side.