A reader writes that Symantec is laying off about 60 people from the division formerly known as Veritas:
By now I’m sure you’ve heard about Symantec downsizing the (mostly) Mountain View storage group, cancelling some projects and moving others to India. One of the fallouts was a soon-to-be-rolled out object-based file system, along the lines of a software version of Panasas. There is a lot of technical detail behind the project but ‘software Panasas for the commodity servers of your choice’ is the 35,000ft view.
Too bad. That file system sounds pretty cool.
Not enough numbers
Looking at their latest 10-Q filing it is clear that their storage business has taken a revenue hit – but it still has the highest margins of any of their businesses. The reason for the revenue drop is the falloff in Sun server sales, where Veritas products are near mandatory.
With Oracle offering its own storage software stack, the suits concluded that this revenue wasn’t coming back. And they’re right about that.
Management is also cutting storage group expenses faster than revenue is dropping. Combine that fact with the shutdown of the new file system it is evident that they have decided to treat the storage group as a cash cow.
The StorageMojo take
I was mystified by Symantec’s acquisition 5 years ago. As I wrote in Veritas goes quietly into that good night:
One of the pay-for-play “analysts” (useful for their role as floaters of trial balloons) was told to opine that mixing security expertise with storage expertise would result in more secure storage, but that is unconvincing. The basic problem with highly secure storage is that systems need low latency access to it, so the opportunity to secure storage through software is fairly limited.
No, Veritas is MBA smart and marketing stupid. Does anyone believe that customers are happy with storage management software today, when 40% of backups fail, data disappears every day, and storage is far from a commodity based on the margins of storage system vendors? This all suggests a massive market opportunity, if only marketing can get their arms around it. But Veritas, despite excellent resources, smart people, great margins and a broad penetration into the F1000, couldn’t figure out what to do to grow their business.
Nor is Symantec likely to provide the missing expertise. Securing corporate PCs against viruses and malware is simply not core expertise for improving storage management.
While they done good work continuing to enhance NetBackup, Symantec is right to retreat from storage. As a company they just don’t have the chops. What happened to the OpenVision sales and marketing crew?
No system vendor today will be as short-sighted as Sun’s McNealy was when he refused to invest in storage software development, forcing the storage group to turn to Veritas for basic functionality and making them a major player. The next Veritas will have to win on a vertical-by-vertical basis.
The good news: the expanding world of massive file-based workflows is wide open to new solutions. We’re at the beginning of the revolution in massive storage – not the end.
Update: Symantec issued a nuanced corporate
confirmation denial. I’m quoting the version published by Chris Mellor in the Reg, as Symantec analyst relations didn’t copy me:
The bottom line is — Symantec is not exiting this [Storage and Availability Management or SAMG] space, and we remain committed to helping our customers who face significant challenges in managing storage growth and ensuring the availability of their critical information. … Symantec is 100 per cent committed to our SAMG product portfolio. None of the SAMG products that we offer today are being discontinued.
As Chris pointed out this is not a denial that they canned the S4 product. Nor is FileStor, which they did release last year, competitive with current scale-out products. In this market if you aren’t moving forward you are moving backward – and Symantec isn’t moving forward. End update.
Courteous comments welcome, of course.
Steve Duplessie said 600; which number is right? Either way, that’s a huge number of engineers.
60 is the number I got from a couple of sources. It may depend on how broadly you define their storage group.
Maybe it is time for Permabit to open a Silicon Valley development group?
Well, then. If they are dropping the product… open source the “Panasas for any commodity server solution of your choice product,” please. Pretty, pretty please.
Didn’t they just release FileStore? (http://www.symantec.com/business/filestore). It seemed like a promising product.
Interesting development – one reason they’re getting out could be that “Panasas for commodity server solution of choice product” isn’t that easy to do (full disclosure here – I’m the new CMO at Panasas so it’s possible I’m slightly biased 🙂 )
Have you looked at GlusterFS at all? It lacks Panasas’s vertical parity but has tons of configurability, including client-side file replication (which means a translator to do ‘file-RAID’ like Panasas is probably doable, if you still want it after full replication). Plus it’s open source, runs on commodity hardware, and you can buy a support contract if you wish.
Jered, 600 was a wild speculation. But it’s more than 60. The number I heard was on the order of 120.
B, “GlusterFS” is an interesting “science project”. It will take a lot of very prudent development investment to get it to the point where it can become a mainstream storage solution. Lustre comes to mind. As far as I know, installing and running Lustre is still like launching the Space Shuttle, with heat shielding tiles falling on your head (loosely quoting the Lustre guys’ description of the process).
Robin, you were right 5 years ago — there was no rational reason for SYMC to buy VRTS. It was the least synergistic of a variety of possible pairings (HPQ, CSCO, IBM, ORCL, SUNW/JAVA). Even if SYMC could have left VRTS alone, ran it as a subsidiary, I don’t think it would have helped. What drove VRTS into irrelevance were the lack of vision and the lack of focus at the key decision-makers’ level.
Where does the “40% of backups fail” statistic come from?
If we were talking a year ago, I’d agree with you, but now I think it might be safe to characterize GlusterFS as having moved beyond the science project phase and into the early portion of the ‘product’ phase. Emphasis on the early, of course, but a product nonetheless. For example, the Minnesota Supercomputing Institute (MSI) is apparently using it on their ‘Itasca’ cluster – an 8600+ core system. Another university I know of is considering its use, too. Whether GlusterFS moves beyond the early product phase into a mature product is another question, but I think it’s past the ‘science project’ phase.
Lustre has gotten better (more stable), but the process can definitely still be difficult for a first-time install. Fortunately it gets easier as time goes on; it’s complex, but not chaotic. When looking at GlusterFS vs. Lustre, it seems the former has much more flexibility (data replication, no kernel modifications, no single metadata server), but less direct throughput for large streaming I/O on the same hardware platform. Each has a role, I think. It’ll be interesting to hear Marc Hamilton’s roadmap for Lustre, promised by the end of the month.
Interesting but not surprising. As a systems and storage admin I’ve had less need for veritas s/w on Sun systems since Sun released Solaris 10. Having a good native multipathing layer built into the OS has alleviated the need for VxDMP which for many of us was the value add that made VxVM worth the price. With ZFS and MPxIO included with Solaris 10 it’s pretty hard to justify the expense of veritas s/w.
I think Netbackup has plenty of life left. If you have a very large and dynamic datacenter then I think it would be hard to get by without netbackup or something similar.
“Where does the “40% of backups fail” statistic come from?”
Real life? I have been involved in a couple of wholesale restores of all existing backups and I can state unequivocally 40% is spot on.
If you think it’s unrealistic, I challenge you – esp. if you are backing up to tape – to restore all of your backups and see just how many really produce something useable. Having done so for enterprise email (not once, but twice!) it’s really eye-opening.
Tape does suck. And I’m even wary of solutions like Copan Systems MAID which attempt to proactively manage backup data (far above most systems out there) and drives for failure. If you are not doing some sort of active management of your backup data, it may not be there when you need it.
As I have others stated, so I can’t take credit for (but wish I could): Backup is one of the most misnamed items in all of IT. It should really be called RESTORE because that’s what’s important!
Veritas Group is nothing. Internal meeting today announced that Symantec are getting out of the consulting and product enablement side pretty much all together except for strategic accounts and instead partners will used to deliver the services instead.
Effective almost imediately what’s more. If a customer calls today and says they want us to do a install or upgrade or consult, Symc say sorry, here is a partners telephone number.
Problem is at least where I’m located there is huge gaps in our partners ability to deliver!
And Management think this is a good thing. They barely mentioned the effect this is going to have on the 11000 employees in Symantec Services Group!