Any idea that Acopia execs took the F5 acquisition money and ran was wide off the mark, according to Rick Gillet, now F5's data systems architect and ex-CTO for Acopia.

Chris Lynch, the old Acopia CEO, and his senior colleagues were all finding that penetrating the large enterprises that Acopia's ARX NAS (network-attached storage) aggregating switch was ideal for was proving difficult. Big enterprises talk to big fish like EMC and NetApp and HP, not to minnows in the storage water like Acopia.

F5 was already talking to enterprises and had a network view of the world. So Acopia decided to take the F5 dollars and ride into enterprises using F5's existing relationships to penetrate the glass ceiling holding it back.

Rick Gillet agreed that Acopia's ARX switch was ideal for enterprises who needed to aggregate multiple NAS boxes. I suggested that it was for enterprises who'd over-spent on NAS but he didn't like that phrase. He also didn't like my suggesting NetApp as Acopia competition. No, Acopia could in fact facilitate NetApp sales. An enterprise could have far too many Windows-based NAS boxes. Acopia could aggregate them and have NetApp introduced and the Windows NAS data transferred to NetApp's filers.

So there's no reason for NetApp to fear Acopia. If I were a NetApp account manager with a good corporate customer and Acopia came in and let that customer aggregate the NetApp boxes with cheaper NAS boxes from elsewhere I would not be pleased, no sirree, not at all.

Rick Gillet agreed that Netapp could aggregate NetApp boxes using ONTAP GX but Acopia does it better. He suggested that, with Acopia, a lot of tier-1 NAS, using say expensive Fibre Channel disks and taking a long time to backup every day, could be split into less tier-1 NAS and a relatively large chunk of tier 2 (SATA disk) NAS with Acopia migrating the data. The result is one logical pool of NAS with only real fast-access data left on a much smaller number of tier 1 disks and the rest of tier 2 disks which didn't need backing up so often. He said one customer, Wiley I think, had more than paid for the Acopia system out of backup tape savings and: "We're not a cheap box you know."

He also said Acopia's IP was in the ARX switch; the box that ran the database containing the metadata about the aggregated NAS box filesystems and that also ran a set of load-balanced network processors interfacing to accessing servers wanting NAS access. and doing database look-ups. There's 2 million lines of Acopia code altogether and nobody else has made a product like his; there is no sign of anybody else doing it either.

The network processors are off-the shelf chips with a real-time O/S but the code they run is Acopia IP. The database is also Acopia IP and it runs on a Linux, PowerPC platform. The actual database itself is stored in a NAS box, any NAS box, connected by Gig E. The ARX switch has a boot disk but that's all the local disk storage it has; it's basically a very fast and scalable look-up system to a database stored on a connected NAS box. The thing scales by adding network processors and/or database processors.

Then Rick Gillet said the natural top size of a NAS box was about a terabyte and a 100 million files or so. Above that point backup takes too long. That seemed odd; you can get a terabyte on a single SATA drive now. Is a natural top-end NAS size now one disk? Disk-based backup and snapshots deal with the tape backup window problem. It seemed a stretch too far.

He discussed flash-based hard drives and said he has a Samsung flash-based notebook. It is one hundred times faster for reads but ten times slower for random writes than a spinning hard drive. He thinks flash is an unbelievable killer technology for reads and said that Acopia's filesystem meta-data database was mostly used for read accesses.

It's not too much of a stretch to think of the possibility of flash solid state disks (SSDs) being used in NAS products once affordablity improves and the write limitations are dealt with.

Sealed lips on F5 purchase matters

I ran into Tim Pitcher, ex-international sales VP for Acopia, and asked him about his leaving Acopia when F5 purchased it and joining Copan. His lips were sealed by a confidentiality agreement. I ran into another ex-Acopian who'd joined FalconStor. Again lips were firmly sealed.

The buzz is that there is a lot to talk about and huge amounts of money were involved; it was a $210 million all-cash purchase after all. Ex-CEO Chris Lynch is no longer directly involved with F5's Acopia business unit; he now oversees F5's data solution technology portfolio.