Last week brought the news that two more storage vendors, Copan and Exanet, are standing with one foot in the grave. It also brought an announcement that Dell was going to resell Qlogic's FCoE CNAs. Journalists and bloggers jumped on the news with some solid analysis and some overreaction. Curmudgeon that I am, I felt I had to throw my two cents in as well.
My friend George Crump correctly called out 3Par, concerning Marc Farley's overstatement that disk was a dead end for archiving. He also correctly identified the first player's burden as a contributing factor in Copan's demise. We've seen several times that the first player spends so much of their limited 15 minutes of fame explaining their technology that when a second player comes along they can say, "We do what Copan does plus we solve your real problems."
Copan failed because they sold a feature not a solution to a real customer problem. For years pundits argued that data deduplication was a feature not a product, while Data Domain, Exagrid, Sepaton and FalconStor quietly sold backup disk targets with dedupe to improve the backup process. Data Domain and the rest took advantage of the fact that backup targets, as virtual tape libraries (VTL), were already a product category. As a result they could make their message, "We have a better backup target than the VTL you were planning on buying anyway."
Copan chose to sell MAID system as effective storage for persistent data solving the power problem most IT folks weren't worried about yet. While they OEMed FalconStor's VTL and dedupe software they never managed to get any VTL mindshare, having spent too long selling MAID technology. What my co-bloggers neglected to say about Copan is that their Revolution product had some issues.
It was big and dense, which here in NY, where data centers may be on the 47th floor, was a problem as the building wasn't engineered to hold 3,000 lbs in that little space. Of course I've had clients have to add steel plates when installing a new UPS, too. However, the real problem was that Copan, drinking their own Kool-Aid made the power supplies too small so that a Revolution couldn't spin up more than 25 percent of its drives at the same time. That's probably fine for a deep archive, but it could really slow down the post process dedupe or ten users retrieving different files at the same time. Once Nexsan, HDS and the rest added spin-down to their arrays it became easy for the competing sales guy to say "Is it worth an extra five percent power savings to risk it?" Add in that they didn't do RAID-6 or equivalent, which I'm on record as calling a must for large drives. As a result, I never could find a place where a Copan Revolution was clearly the right solution.
