HDS validates FalconStor dedupe

I’ve know for a while that FalconStor is one of a few vendors to have truly multi-node, global dedupe.  They’re on a short list with SEPATON (8 nodes), Exagrid (10 nodes), and NEC (55 nodes).  (Data Domain is getting there, but their GDA is only available with the 880, NetBackup and OST, and only uses two nodes.  IBM has had a 2-node cluster for a few years now, but has been unable to add a third or fourth node yet.)  Quantum has no multi-node option.  I believe that HDS understood the value of multi-node dedupe for their customers and wanted to pick a product that offered that feature. I’ve written a lot about this feature here, here, and here.

But when I would talk to FalconStor’s competitors, I always heard things like, “their stuff doesn’t work.”  Which is a surprising statement about a company that sold more VTLs than anyone else in the world (most via OEM agreements via EMC, IBM, and Sun).  Just last week a competitor tried to tell me that FalconStor’s box didn’t actually do real dedupe; it only did object-level single instance.  (This comes from FalconStor naming their product Single Instance Repository, although it really does do “real” dedupe.)

They haven’t had the greatest time financially lately, with two bad quarters in a row.  The stock price doesn’t look too good either.  The biggest problem is the loss of their OEM revenue.  Once the darling of the VTL industry, Falconstor has seen their OEMs go their own way, with IBM acquiring Diligent, EMC acquiring Data Domain, and Oracle cancelling the Sun deal.

So I’ve been quietly watching as HDS was trying to pick another dedupe partner.  They were a Diligent partner then IBM bought it.  Rumor has it they were going to be a Data Domain partner before EMC scooped them up.  So they’ve been quietly looking at their options, and I’ll give my thoughts on how they ended up where they ended up.  (These are my thoughts based on conjecture, not any actual knowledge of their internal meetings.)

Exagrid would have been a contender, but Exagrid is aimed at the mid-market and HDS wants to aim for the enterprise.  (They have no Fibre Channel interface, for one thing.)  SEPATON is aimed at the enterprise, but their content-aware dedupe isn’t “aware” of CommVault yet.  (They can store CommVault, but they can’t dedupe it yet.)  Since HDS resells CommVault, that pretty much killed SEPATON for them.  If they did consider NEC, it might have been seen as a competitor.  NEC also still doesn’t have a Fibre Channel interface, which the enteprise needs.

That left Falconstor as the only viable vendor to offer multi-node dedupe, a Fibre Channel interface (and a NAS interface), and that handled any backup type.  But you’d better believe that HDS beat the crap out of the thing before they decided to risk their name by reselling it.

Which brings me back to the title of this blog post.  I’ve known that Chi Corporation has been reselling the crap out of the FalconStor dedupe stuff (word is they’re FalconStor’s biggest reseller).  But there’s nothing like having a company like HDS putting their name behind your product.  By agreeing to officially resell FalconStor, I believe that HDS has silenced FalconStor’s dedupe critics forever.

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

3 comments
  • Up to 16 nodes I believe… For a total of 196TB’s of deduped data that can be managed within a PureDisk pool.

    Personally I’m holding out for NBU7 to get global dedupe between media servers.

  • As far as I know, all source dedupe products have global dedupe, and Puredisk falls under that umbrella. I know that you can use it in a target dedupe capacity, but until I actually see customers DOING that, I will continue to forget that is the case.

    As to NBU 7, you mean you’re waiting for them to ADD that feature to it, right? Because it wasn’t in the shipping version.