Data Domain now says Post-Process is OK

In a surprise reverse of their longstanding “inline is the ONLY way to go and anyone that does post processing is stupid” position, Data Domain today announced that Post Process is A-OK with them.  Or at least that’s how I’m reading the acquisition of Data Domain for $1.5B by NetApp.  Click Read More to see more.

NetApp announced on a concall today that they are acquiring Data Domain for $1.5B.  Nice multiple.  That’ll be nice for all the other dedupe players who are waiting for IPO or acquisition.

But I envision some significant changes in future backup school presentations by “A Damnation Tapped.”  (That’s the best anagram I could find that includes both companies’ names.)  ASIS (Advanced Single Instance Storage: dedupe on primary data on a filer) & NetApp’s NearStore VTL both do post-process dedupe.  Data Domain does inline dedupe. I can’t see “A Damnation Tapped” saying in their sales presentations that post-process is bad.  Therefore, they’ll be saying that it’s OK, which was the point of my blog entry.  (It made a nice headline anyway.)

Also, I question the long-term viability of the NearStore VTL.  (I don’t think ASIS is going anywhere.) Once the acquisition is complete, NetApp will have two products that both backup data at similar speeds and dedupe said data.  One does replication and has no back-end tape, and the other has back-end tape and no replication.  Both limitations can be addressed with a few months of dedupe.  Surely they won’t fix the limitations in both.  Surely they won’t dump the Data Domain product they just acquired.  So what’s left?  The NearStore VTL.  That’s where I’m at.  I know that right now there are reasons to choose NearStore VTL over Data Domain, but they can close those gaps with development.  Once they do that, I think that product is going bye-bye.

As to ASIS, it’s the only product that is deduping VMware datastores with little or no performance impact — and it’s included with WAFL at no extra charge.  That’s going nowhere.  Even if it does nothing but that, it’s going nowhere.

Written by W. Curtis Preston (@wcpreston), four-time O'Reilly author, and host of The Backup Wrap-up podcast. I am now the Technology Evangelist at Sullivan Strickler, which helps companies manage their legacy data

13 comments
  • You do like to try and stir the pot, lol; putting words in others’ mouth.

    Sounds like a pretty nice problem for NTAP to have; it’s something called OPTIONS, rather than force-selling whatever you have in your sales bag, and customers tend to like having OPTIONS. “Oh, you want a VTL-tape based solution? Ok we have that.” “You’ve changed your mind and want post-process instead of inline? No problem; we can provide that technology if you prefer”.
    We’ll have to wait and see how NTAP decides to go to market with this, but they have some nice options to pick from; again, a nice problem to have!

    Far from the cobbled-together solution that EMC is trying to sell with one vendor’s dedupe on top of another vendor’s VTL; YIKES!

    BTW…..how often do you get told you have quite the resemblance to Newman? (Helloooooo Newman!)

  • Several points in favor of moving quickly to put the legacy NTAP VTL platform to rest, in favor of the DDUP VTL platform:

    1. The existing NTAP VTL models (VTL300, VTL700, VTL1400) can all be converted to regular NetApp heads, simply by loading Data OnTAP instead of the VTL-OS. Not for free, unfortunately, but at least you are not stuck with a useless lump of unsupported VTL controller and disks. Now I know why they were harping on “investment protection” so much in their sales pitch ๐Ÿ˜€ As far as the older original Alacritus stuff goes, if any of it is still out there, I guess you are SOL, but that has been EOL for quite some time.

    2. The price NTAP paid for Alacritus was only $11 million back in 2005, so they are only losing that amount, plus whatever resources invested since then on the NTAP VTL platform. Vs. the $1.5 billion for DDUP, it’s probably chump change. A good portion of that also can likely be salvaged by porting critical features to the DDUP platform. Personally I would love to see tape export from DDUP (either reinflated or still-deduped), but don’t expect that to happen any time soon.

    3. No idea what the existing market share of the NTAP VTL was, but surely it is dwarfed by the DDUP VTL market share. The stated plan is to take share from EMC and others, so if they have to lose a point or two short term to grow long term, I don’t see that bothering them much. Also note NTAP said IBM could resell DDUP now. I wonder if Big Blue will take them up on it despite their fairly recent Diligent purchase. ๐Ÿ˜› Otherwise this could be a point of conflict down the road between the partners.

  • I like idea #1. Not for free? If they do indeed kill the product, it should be for free, IMHO.

    I would also like to see tape export out of a Data Domain box. I hope, however, that it’s done in a way that supports tape stacking with the media manager still knowing what’s where. I don’t like the one-bar-code-to-one-tape relationship thing. Never have. I’d also like to see a deduped backup to tape just to back up the DD box.

    Not sure what this means to the IBM relationship.

  • I would like to offer another interesting point of view:
    Let’s say that an inline dedup company builds a very easy to use NAS appliance with a very good dedup ratio, which has snapshots and even fiber connection. Now let’s think of a market whose data is exploding and understands the great value of DDUP.
    I assume you get my point – ASIS is nice – but DEDUP can be great especially in a VM env.

  • They are two different tools to do different things. ASIS does primary dedupe, and DD does secondary/backup dedupe. Neither does the other well.

  • I agree, the aquisition of DD has got to herald the end of the NearStore product. Choice and options for the customer is fine but confusion of the message is a nail in the coffin of any deal. I don’t think it will happen though until they have done some dev work on the DD to get that oh so important feature of clustering and replication between boxes.

    Another thing netapp may want to work on is getting the DD to actually scale to some useful size for enterprise customers.

    In terms of the relationship between IBM and Netapp (which can sometimes be a strange one to understand) I would be interested to see if the DD boxes appear in the Nseries range (currently the NearStore VTL doesn’t get OEM’ed by IBM). I am sure all will become clear…

  • No, they don’t have clustering, but they do have replication. As to clustering, that will probably go hand-in-hand with their plans for scaling.

    I highly doubt we’ll see DD boxes being sold by IBM. They have their own inline dedupe system.

  • Now EMC wants to muscle in. Serious play- or are they just trying to make NetApp bleed cash as a competitive move?

  • [quote name=W. Curtis Preston]I highly doubt we’ll see DD boxes being sold by IBM. They have their own inline dedupe system.[/quote]
    Curtis – do you see IBM wanting to get into this now that EMC is there, to deny the technology from ending up in “the wrong hands”?

  • Not to imply that EMC isn’t serious, but for them there isn’t much of a downside to a bidding war. Even if NetApp wins, EMC can drive the price up to a point where winning may hurt too much. If EMC wins, then NetApp is damaged in the market, and has trouble with the “our de-dupe is fine” story. In the end, EMC has the pockets to decide exactly what the price is, even if they aren’t the ones who pay it. I think they’d be happy to win, but losing hurts them less than it hurts NetApp- and NetApp could be hurt more by winning too.

  • Both companies have told the world with their bids that the product they have isn’t meeting their customers’ needs. EMC’s first generation DL3D product has serious challenges in scalability and restore performance. NetApp’s first generation product (for the backup space) doesn’t yet have replication or RAID6 and also doesn’t scale.

    Whoever loses is screwed.