I believe he considers EMC to be naive when it comes to NAS.
I believe Michael Ruettgers is on record as saying that he does not agree with the way NTAP implements NAS. It looks like EMC will continue to position Symmetrix at the core with current and future versions of Clariion defining different price/performance points all the way to the edge. Having competed with the likes of IBM and SUNW at the highest performance segments of the storage market while creating close to $150 billion worth of shareholders' value over a 10 year period, I would be more inclined to take his word at face value.
With his McData (more below) and Aviion spin-off chips in hand to go with the strong growth of the core business as well as prime positions in the architecture of the Intelligent Storage Network tightly integrated with the next-generation Intelligent Network, I daresay that he might be just be understating EMC's total return prospects (including spinoffs) over the next 10 years after making the stock appreciate 81,000% during the 90s. Richard Egan, the founder, also has a venture capital company that will probably not be shy about exploiting the storage and non-storage opportunities that will be available as the Intelligent Network is architected. Storage is the nexus of that network.
I know of no other "thin server" on the market except NTAP.
You probably mean that there is no other "thin server" exactly like NTAP's filers. Well, that what everybody says about the products of their favorite companies.<g>
As defined in that 1995 NTAP paper that you linked, a storage appliance is one that is designed to perform only one function very well. There are obviously many implementations of that concept, yes?
Nor do I know of any trend to consolidate thin servers. The trend that NTAP sees and is promoting is the consolidation of "fat servers".
This one's a question of semantics. A NTAP filer or thin storage server, which is designed to perform only one function, is thinner than a general-purpose UNIX or NT server, which is designed to perform many functions, which in turn, is going to be thinner than the next-generation general-purpose UNIX or NT server etc.
The old reliable tape drives, housed in autoloaders and libraries, are probably the archetypes for the storage appliance because those were designed to perform back-up and restore functions only. Look it up and see the way they consolidated over the years to see what I mean about technology consolidation patterns in corporations
You're probably not suggesting that stringing together 100 or more 1.4TB NTAP filers all over an organization with multiple locations is not going to create all sorts of inventory and storage management problems now, are you?
These concepts are Darwinian....
..... I am sure that NTAP and EMC are both contributing to SGT's profitability and product direction.
Take a look at this.
xiotech.com
Seagate paid $330m for a SAN vendor like Xiotech late last year on the heels of NAS acquisitions by Quantum, WDC and Maxtor. Maxtor and WDC, by the way, both decided to pull the plug on their enterprise drive divisions and decided to be their own customers in the NAS space. The main point is that they are already competing with NTAP at the low-end. That's the structural change in the DD industry to watch because it directly affects NTAP's revenue base. That's the structural change to watch because consolidation will
only occur around these disk drive companies, not without them.
EMC assembles all its components, including customized firmware and disk drives. It manages its supply chain well and plugs its key suppliers into the feedback loop from its customers and that holds true for its key disk drive suppliers.
There are two basic ways to go with the disk drives: optimize speed or optimize capacity.
In terms of sheer capacity, the Elite was the former capacity champion until it was displaced by the new Cheetahs. EMC and Seagate collaborated on the Elite program -- which uses an industry-leading 14 3.5" platters on a disk drive spindle running at a slow 5400 rpm. The engineering required to make that kind of drive is not trivial and only Seagate offered this customized drive (currently at 48 GB).
In terms of speed, Seagate was first with the 7200 rpm Barracuda, the 10,000 rpm Cheetahs and the 15,000 rpm Cheetahs, which incidentally uses 12 2" platters on a spindle. These high-performance drives tend to rollout slowly and be in short supply while being premium priced for extended periods of time because of Seagate's technology and logistical lead. IBM, for example, lost an estimated $350 million in lost business in drives and arrays when its 10,000 rpm drives encountered manufacturing problems during the last two quarters.
So, not only does NTAP have to contend with competing with Seagate's Xiotech at the low-end and the middle market, it also has to compete for those high-performance drives as they rollout slowly.
I contend that NTAP's NAS is a discontinuous innovation. It is discontinuous not because it is NAS, but because the WAFL system and all of its inherent efficiencies and the software that NTAP has built around it offers quantum leaps in price/performance/reliability and total cost of ownership versus UNIX and Windows file systems. NTAP customers are literally replacing their UNIX and Windows file systems with WAFL, turning their UNIX and Windows servers into more efficient app servers.
I disagree. WAFL is the file system of a storage appliance designed to perform only one function very well. As such, it is not designed from the ground up to perform a host of other functions required by a corporation. Those functions will only change as the network becomes faster.
For example, it is true that by using 'copy-on-write' one can use Snapshot to create up to 20 copies of the data, but those are READ-ONLY copies. In comparison, EMC's TimeFinder can create up to 8 production copies or BCV (Business Continuation Volumes) -- READ and WRITE copies -- that can be used for other tasks with no performance penalties for frequent database resynchronizations. That enables backup, analysis, testing and SDRF which allows the movement of massive amounts of data over long distances.
NTAP customers are literally replacing their UNIX and Windows file systems with WAFL, turning their UNIX and Windows servers into more efficient app servers.
I disagaree with your characterization that implies that there is a wholesale move to WAFL over Unix and Window servers. The 1999 NAS market was less than $900 million and it is expected to reach $1.2 billion this year. NTAP had 60% of this niche market, but EMC's Celerra grew faster from $7 million in the 1Q98 quarter to $90 million in the 1Q00 quarter and EMC is clearly the vendor to watch because Clariion allows it to scale from the dominant Symmetrix at the core to highly specialized Clariions at the middle-market edge. Note that EMC currently in the process of spending $2.5 billion in R&D (75-80% on software) over the next 2 years to address the segmentation in the booming storage networking market. EMC is basically plowing back all its software profits (over $1.2B annualized based on 1Q00) to continue to widen its 3-year lead over the competition.
Here are some datapoints that indicate the early stage growth of storage networking:
This is from a March 20, 2000 IDC report:
For server vendors, appliance servers are too hot not to touch. IDC estimates this quickly emerging market will develop into a $11 billion-plus opportunity worldwide by 2004 - this from a market that was significantly under $1 billion in 1999......
.......According to IDC, network attached storage will represent the largest opportunity for appliance servers, accounting for more than half the market's value through 2004. Revenues in the Web server segment, however, will increase the fastest. The compound annual growth rate (CAGR) in this segment will be more than 130%, compared with a 73% CAGR for the overall market.
idc.com
That kind of torrid growth is paralleled by the equally torrid growth of McData, which, not surprisingly, has 99% of the current Directors' market largely due to EMC's dominance at the sites requiring high-performance.
......According to IDC, Brocade was the market's number-one supplier in 1999, with 39% of revenues. "Brocade's focus on fabric switches positions it well overall and gives the company a technology base that could be applied in the future to the lucrative director class switches," Gray said.
McData was second with 22%, and Gadzoox rounded out the top three with 17%. Vixel is the top ranked unit supplier with a 48% share of the hub and switch market. Vixel was also the leading port supplier in 1999.
Director switches and loop switches will be the fastest-growing areas. Director switches will increase at a CAGR of 129%. They will overtake fabric switches in 2001 in terms of revenues, and by 2003, they will account for 52% of the market's revenues.
"Fabric switches have become more affordable as prices declined rapidly. In contrast director switches have experienced minimal pricing declines, and because of the product?s high-end, support-oriented nature, we don?t expect rapid price declines in this segment. As customers focus on more high-availability installations, we expect directors will take an increasing share of revenue." Gray said.
idc.com |
