Watch it play out – again – in storage
This morning’s Wall Street Journal has an article about startups – I won’t link to it because subscription required – where they note that a number of real companies have bought from startups lately.
Why?
Now corporate tech managers are once again starting to buy equipment from small networking businesses with little-known names such as Riverbed, Aruba Networks Inc., Isilon Systems Inc. and BigBand Networks Inc. Many of these small firms make products that solve new corporate-technology problems, such as how to most efficiently store new corporate data like video, or how to best improve the transmission of information across networks that have been weighed down by multimedia applications.
While some of the big tech firms also offer similar technology to deal with such issues, tech managers are finding that the start-ups often have more cutting-edge products that are cheaper than the big suppliers’ offerings. “The old guard equipment guys are having to think about more than just equipment [and] they’re having to think about software and video,” says Joe Skorupa, a research analyst with Gartner Inc. “They aren’t used to thinking in those terms.”
Multi-media apps, eh?
That is just where Isilon has built its business, right where NetApp should have owned the market. As NetApp’s marketing VP Jay Kidd recently noted
I wish that we were big enough to invest in every single segment. About three years ago we really targeted the enterprise segment and wanted to build a business there, and we had presence in the technical segments. Did we leave some segments unattended? Yeah, we probably did. And Isilon has gone and found those segments and called our attention to them.
It’s a challenge to balance, because even though we’re growing 35% a year, we can’t invest in everything we’d like to. We’re doing more things right now — we’ve broadened the focus in the past year or so to be back on the technical side again.
Five years ago plenty of startups were going expensively broke going after “new media” markets, usually employing the costly strategies of “get big fast” and the like. Isilon stayed lean and focused. Now they, and not NetApp or EMC, have the brand and the references for media apps and scalable cluster storage.
Jay’s comment is right on: NetApp doesn’t have enough money to invest in everything, so they invested in the enterprise. Wall Street is loving that decision – and so is Isilon. The new niche markets are never big enough to attract the leaders. By the time they are, the little and nimble competitors are the incumbents.
From little acorns do mighty oaks grow
Or something like that. Not everyone can be as lucky as EMC was in the mid-90’s and confront a comatose IBM (IBM has yet to recover culturally from that upset). Frontal assaults on much larger entrenched competitors are great fun for the first couple of months. Isilon has been both lucky and smart.
The StorageMojo take
The secular trend to cooler data and a “keep everything” mentality is fundamentally reshaping the mass storage market. Many new niches are opening up to software focused storage companies. Isilon is one of the first. It won’t be the last.
Comments welcome, as always.
Good morning Robin:
I found the WSJ article interesting but I am left wondering where can you get a formula that objectively evaluates Personnel, Price, Performance, and Power consumption for a storage subsystem?
Objectively evaluating storage equipment is a seemingly impossible task, however, I can assure you that my customers would be excited if such a tool exists. Conversely, I am pretty certain that the data storage manufacturers would be very unhappy unhappy if there is an objective storage tool out there. Perhaps the decision to buy Isilon was based more on price and sales pitch then on any objective storage performance and objective operating cost analysis.
Mike,
Great question. Like you I’d love to see that objective formula.
That formula has to take into account two major variables: goals and values. As I noted in Risk perception in data centers people do a terrible job of evaluating risk, which distorts their goals. They tend to focus on the calamitous but unlikely while ignoring the every day risks that are much more like to bite them.
Values are equally fluid. As a result, a decision that feels great on Monday can look pretty bad by Friday.
That said, I do believe that customers have come to recognize that manageability is a valuable feature and now rate it higher than they did 10 years ago. That is one reason that NAS is continuing to grow in popularity.
Robin
Riverbed, Isilon, and Aruba Networks are “little known”? To whom? I’m surprised the WSJ would so easily rely on baseless (or at least unattributed) conjecture. Their articles are typically better than that (but don’t get me started on their editorials).
Isilon’s story is pretty interesting, but I think it has less to do with Netapp’s inability to cover all the bases than it does with SGI exiting a market and leaving it without a replacement. SGI used to own media, and they still have a business there, but they have fallen considerably and have suffered from a lack of market definition and more affordable solutions for media customers
At the same time, video integration is now on the rise as a corporate IT technology. Isilon stepped into a market niche that was once seen as being too limited to worry about by most enterprise vendors, but that niche is now taking off. The question is whether or not Isilon will be able to find a sufficient market that aligns with its capabilities and value proposition. Don’t get me wrong, I’m not being a doubter here – Isilon has done very well. For me, the question is what impact media will have on corporate IT. It could be significant as Google is leading the way towards increased video content as part of its core search engine. Look at Technorati – they are listing videos in their searches too. Corporate video is coming on and Isilon may have the advantage they need to stick.
Karl,
It’s all relative to the HPs and IBMs of the world.
Marc,
I liked the SGI dropping out story until I remembered that SGI owned the special effects business in Hollywood for a while. My sense is that Isilon, while they certainly are selling in Hollywood, have actually opened up a new market, massive multi-media content storage. Maybe SGI would have grown into it had they stuck around (well, actually they are still around) but only if they offered a cluster storage option like Isilon’s.
I do agree that online video is a new frontier for corporations. With the plunging cost of production and the acceptance of YouTube quality video there is no reason that text shouldn’t be augmented by video shorts.
Robin
RE: “I’d love to see that objective formula”
I wonder if LeRoy Budnik of Knowledge Transfer has done this?
He would be the person most likely and he has something pretty close.
Maybe he will read this post and reply.
Keep everything.
Forever.
Especially interesting if you are both SOX and FDA regulated. I would not mind seeing your take on Exagrid.
Their byte level ‘de-duplication’ is WAY ahead of the standard block level stuff so many are handwaving about right now. Only time will tell, but they look like they could be another Isilon in that sector. I know we are looking hard at them right now.
Mr. Stay:
According to http://www.exagrid.com/resources/ExagridByteLevelWhitepaper.pdf Exagrid doesn’t do de-duplication, they do file level (it is strongly implied) “Byte-Level-Delta Data Reduction”. I.e. they are a wrapper around the FOSS rdiff-backup concept, which goes back to RCS at minimum.
This will do well for some applications, rather poorly if you are e.g. backing up a zillion machines where the vast majority of the data is the same system and application binaries, and their strawman attack on block level de-duplication fails to impress.
I do not believe the ‘keep everything’ corporate mentality generally refers to individual PC system and application binaries. (This is not to imply there are no occasions when that type of backup is not critical to some types of business.) However, for the ever growing central repositories of unstructured data Exagrid does appear to do an excellent job and it matters not at all how old the underlying data structures and algorithms may be.
“I do not believe the ‘keep everything’ corporate mentality generally refers to individual PC system and application binaries.”
Well put—in fact, here we’re really talking about two very different things, archiving for “keep everything” (since among other things, easy and sophisticated access afterwards is required to reasonably satisfy retrieval requests), and backup for continuity of operations and all that.
Exagrid might indeed be a good starting solution for the former—but that’s not what they’re selling (according to their web site), and I didn’t see any of the archiving features that you’d desperately want for “keep everything”. They’re selling their stuff as a pure backup product, with compression of even the most recent copy of your data. I could build a very good archiving system on top of that, but it would be an addition.
I’ll also add that in operational terms, these have to be complimentary solutions, and it’s not clear to me that for either a segregation of types of data is a good idea. Not only does that tend to get you into trouble by inadvertent omissions, which you can’t afford for either, but who’s to say that an archival retrieval would never need to know exactly which version of a particular Windows binary file existed on a particular workstation at a particular time?
Perhaps we might try grounding this in a real case where we know at surface level some of the technical details, AMD’s antitrust case against Intel, where Intel is saying “The dog ate my email….”
Kevin and Harold,
Love the point about the binaries. How much data is stored in application specific formats that are unreadable/unsearchable or unformatted without the application?
For example, email formats. Various mbox formats are “standards” for Unix mailboxes. Apple Mail used to use the mbox format, but with the move to Spotlight, they needed individual files to index. So no more mbox. You can save emails into the mbox format (Save As> Raw Message Source) but how many people even know there is a standard mailbox format? If you don’t, is Apple Mail only readable by Apple Mail?
And Exchange is in its own place entirely.
Interesting issue. Thoughts?
Robin
RE: “I’d love to see that objective formulaâ€
Read LeRoy Budnik’s comment at:
“Go Get ‘Em, Bruce!”
http://www.drunkendata.com/?p=1230
[excerpt]
“Think of the combined electronics load of the array, switching, cooling, etc as 10W/drive. You would not believe how much adding encryption adds to the load. Then consider the problem for HVAC. I have been working out the formulas, it is not pretty.”
Robert Pearson comment:
I have been working on the iSCSI “Speed Limit of the Information Universe” numbers. A corollary to throughput is the “green” cost. Is it “greener” to write slower? How slow is too slow?