It has been a rough and tough 3 weeks and I missed writing my blog. Last week, the toughest of the 3, was my CompTIA Storage+ training to Symantec SEs in Malaysia. They were a great crowd, and I loved it but I was really tired after that.
One exciting news during that week was the ouster of long time employee, and CEO of Symantec, Enrique Salem and replacing him with Steve Bennett, their Chairman. The news of that unfortunate event can be read from here and here. And almost hours after that, the calls to break up the Veritas portion of Symantec came up and putting pressure on the board of directors in Symantec to either spin-off the entity or sell it off.
To be fair, many observers, including me, believed that the marriage between Symantec and Veritas in 2005 wasn’t really what you would call a “match made in heaven”. It was more like strange bedfellows to me. And there was an internal joke (one that I could not verify) about the Veritas CEO, Gary Bloom’s promise to the Veritas board when he joined them from Oracle in 2000.
It went like this:
“Gary Bloom promised the Veritas board of directors in 2000 that he would be able to bring Veritas to a USD$5 billion dollar company in 5 years time. Nearing the end of the 5 years in 2005, Gary fulfilled his promise by merging with Symantec, instantly making Veritas a USD$5 billion dollar company.”
Note: This is just an inside joke which I heard from a Veritas friend back in 2005, and by no means put Gary Bloom in a bad light. If I did, I apologize.
But back to the present. Our class last week brought up the subject of Symantec FileStore. When it first came out in October 2009, I thought it was an interesting solution. For once, I thought there was something could “out filesystem” NetApp’s ONTAP and WAFL, because Veritas had one of the best scale-out, clustered file systems. They just haven’t figured out the front end protocols yet, where NAS and iSCSI reigned. Veritas File System (VxFS) and Veritas Cluster File System as part of Veritas Cluster Server (VCS) was mature and proven in the enterprise. Along with Veritas Volume Manager (VxVM), this was perhaps THE best file system/volume management suite around. Mind you, ZFS hasn’t reached the level of prominence yet at that time.
Filestore is a scale-out architecture up to 16 nodes, and its clustered file system supported up to 2Petabytes. The clustered file system is a symmetric SAN file system, and from the picture above, took any storage volumes (RAID or JBODs) via Fibre Channel, iSCSI or even SAS in the backend, and presented to the clients via NFS, CIFS, HTTP, and FTP. The Filestore software ran on the Linux platform, and it has native a active-active high availability HA architecture. It already has global namespace, with its roots back in VxFS and since it ran on Linux, Veritas, I believe at the time, also had NetBackup Advanced Client on that platform. This meant that any of the 16 nodes can be assigned as a Media Server for data protection reasons. This also meant that Veritas could have enhanced Filestore and the Linux platform to include things like snapshots, data deduplication, compression, cloning, replication (Veritas Volume Replicator or VVR integration would have been a piece of cake) and not to mention, the Symantec security porfolio of solutions as well. And I think Symantec end point protection, the anti-virus software was also integrated as well.
Wow, I could go on and on …. salivating what Veritas (I mean Symantec) could have done. The possibilities are endless, almost.
I recalled that Symantec even went all the way to get SPECSfs benchmarks, although I could not find the results to verify this. There was much potential, don’t you think?
But what happened to Filestore now? I didn’t ask the Symantec students much about it and I am not sure what the roadmap is for Filestore.
This recent news of Symantec being pressured to “do something” with their Veritas legacy gives me a good opportunity to give my 2cents worth of advice to Symantec. Let Filestore run free … Open Source it!
The time is ripe for an open, non-proprietary storage hardware platform. For far too long, enterprise storage customers are being enslaved by the very nature of storage array hardware. It’s proprietary, and you can’t mix and match the way you want it to. When you run out of memory, you basically have to forklift the storage array controller just to increase the RAM memory. If the CPU runs out of juice, you are again toasted to upgrade the storage head. Sure, there are plenty of terms such as data-in-place upgrade and online upgrade or whatever, but the truth is, customers are being suckered to opening their cheque books to go to the next level.
ZFS has already been out there with the Illumos Project and it is gaining a lot of ground with many developers. It is beginning to create a whole new generation of enterprise storage solution based on standard x86 architecture. Symantec (or perhaps the Veritasians) should start considering the same, putting Filestore codes out on open source.
And given its legacy of Veritas, it could probably give ZFS a run for its money, because there is much to gain for Symantec by letting it go.
Note: Note that I keep using Veritas instead of Symantec. I am a nostalgic fella, so Veritas it is for me.