The storage industry is fortunate enough to have pretty good scorekeeper -- IDC. They track revenue and capacity by vendor across several categories and geographies.
Every quarter, they use publicly available information to infer who sold what in what segment. And if you do a lot of OEM business, well, that revenue goes to the company that sold it, not the company that made it.
Now, like anything else, IDC is not perfect. Sometimes they infer something wrong, and then there's the endless vendor quibbling about categories -- but generally speaking, they're right much more often than otherwise.
And if you're in the heart of the storage business, you spend a lot of time looking at IDC data.
Why do I think this is interesting?
Because it shows where customers are putting their most valuable asset -- information. It's a trust thing, in my mind.
And -- strategically -- if you believe that information infrastructure starts with where information lives (e.g. storage), well, then, there are long-term implications down the road for who sold what, aren't there?
So let's dive in and see what happened in Q4 of this year.
The first hurdle you'll need to clear is understanding the IDC categories.
- external disk storage (anything outside the server)
- external RAID (subset of above with some sort of protection scheme)
- networked storage (mostly a subset of above, anything not directly connected a la DAS)
- open SAN (subset of above with any block protocol, e.g. FC and iSCSI, but not things like mainframe and NAS)
- iSCSI SAN (subset of above)
- NAS (technically, anything not SAN, includes things like Centera)
Now, if you know about what vendors sell what, you can easily derive additional categories, like
- high-end storage (EMC DMX, HDS Tagma and OEMs, and IBM DSS if you're feeling charitable)
- mid-tier storage (EMC CX, NetApp, part of the HP, IBM, Sun portfolio, and so on).
And, of course, you can do various geo splits, etc.
Now, most of us on the vendor side are more interested in who built the product rather than who sold it, but it's also easy to correct for that, e.g. EMC product sold by Dell, NetApp product sold by IBM, etc.
One more hurdle: the revenue vs. capacity tradeoff.
IDC reports both, but there's an interesting interplay, given that there's a wide range of price points in the storage market.
And then the vendors get the numbers. They privately do a lot of analysis, but what most of us see is the spinning exercise that occurs when every vendor dissects the data and tries to show their view of it in the most favorable manner.
I've written about this before. I don't think it's harmful, but it is interesting to watch.
Somewhere in the footnotes, you'll see that IDC took EMC's capacity numbers way up, and restated retroactively for quite a while back. I guess they realized we were selling far more capacity than they had calculated -- in addition to the strong revenue showings.
And EMC Did Well
No surprise, EMC did very well in Q4, on many fronts.
For every category that IDC tracks, EMC/Dell came in #1, with the exception of mainframe, where we are #2 behind IBM.
If we look at it geographically, EMC/Dell has a very strong #1 position in the US (37.6%), as well as in Western Europe (26.8%), Asia Pacific and "Rest of World" (28.4% and 33.5% respectively), and not so good in Japan.
If we calculate the high-end of the market, EMC is strong with 41.8%. Even higher if you exclude the IBM DSS, which many of us believe doesn't belong in that category.
And contrary to some people's perception, EMC/Dell is #1 in both NAS and iSCSI.
Not a bad picture, to be sure.
But What Else Is In The Numbers?
Once again, high-end storage showed big growth in Q4. A whopping 20.3% growth compared to Q3, where it also grew substantially. There's a big shift to big iron in storage, and the numbers continue to show it.
And then there's iSCSI.
Now, the last time I wrote about iSCSI, I was excommunicated from the iSCSI Fanboy Club (see here), but -- once again -- although there's some good growth off a very small base, according to IDC iSCSI accounted for a mere 3.1% of the overall external storage market. To make matters worse, the growth rate seemed to slack off a bunch from Q3. By comparison, mainframe storage looks like a robust market.
So what's happening here? Any iSCSI fan want to comment?
Is Storage Just A Commodity?
Yes and no. There are lots of choices from lots of vendors, which might be a sign of a commoditized market. And all storage vendors use basically the same parts bin of processors, disks, memory chips and the like -- so many of the ingredients are commoditized.
But I would argue that when you start lining vendors up and look at an extended set of criteria (other than price per pound) strong differences emerge in capabilities, strategies, breadth and value-add. No one that I've met that takes their information seriously thinks of storage as just a commodity -- although everyone is looking to save money.
I've also come to believe a few non-obvious things about the storage market.
For many customers, it's a trust thing. They're trusting the vendor to store their most important asset -- information -- on storage devices. There are many opportunities for a bad day in IT, but I would argue that there's nothing quite like a bad storage day.
And more of the value is shifting to what a vendor can do with storage, rather than the storage itself. Can you tier it? Manage it? Use less of it? Move service levels around on the fly?
There are those that would argue that one-size-fits-all in storage architectures. I just don't get it -- there's so much diversity out there in use cases and vendor offerings, I'd be hard-pressed to come up with an idealized architecture that does most of what people want most of the time.
I also don't believe in one-size-fits-all for health care, automobiles and real estate either.
And, here comes the EMC love: I think we've done an excellent job in building, delivering and supporting a broad range of storage technologies that do the job and do it well. I look at the IDC numbers and think we're doing OK.
But there's always more to do, isn't there?