The G&K thread seems to believe that NTAP, as a young growing strong company, may endanger EMC's great performance.
The NTAP CEO has some sort of grudge against EMC because he approached EMC in 1997 to propose an alliance interlocking NTAP's storage appliances targeting the low-end and the middle market with EMC's high-end offerings. He was rebuffed because EMC apparently had other plans (DG, Seagate/Fujitsu) but his constant references to EMC's arrogant behavior at that time seem to indicate some lingering bad blood which colors the PR.
That bit of color is a prelude to the answer to the question: is NTAP's technology a 'discontinuous innovation?'
It is not.
A thin server is a thin server is a thin server that will eventually be consolidated. No magical file system (WAFL) will change that pattern of usage in corporations. Also, NTAP uses a souped-up version of RAID 4, which has its distinct advantages and disadvantages.
RAID-4 (Level 4)
In RAID-4, parity is interleaved at the sector or transfer level. As with RAID-3, a single drive is used to store redundant data using a parity check byte for each data stripe. Parallel data paths and sector or block level distribution across the data drives allows for independent drive operations and multiple I/O operations to execute in parallel. RAID-4 is identical to RAID-3 except that large stripes are used, so that records can be read from any individual drive in the array apart from the parity drive. This allows read operations to be overlapped.
RAID-4 offers high read performance and good write performance. RAID-4 is a general solution, especially where the ratio of reads to writes is high. This makes RAID-4 a good choice for small block transfers, which are typical for transaction processing applications. Write performance is slow because the parity drive has to be written for each data write. Thus the parity drive becomes a performance bottleneck when multiple parity write I/Os are required. In this instance, RAID-5 is a better solution because parity information is spiralled across all available disk drives. RAID-4 systems are almost never implemented mainly because it offers no significant advantages over RAID-5.
Advantages:
Good data availability
High performance for read operations
Cost effective - only 1 extra disk is required for parity
Disadvantages:
Poor write performance
Poor small, random I/O performance
ctnews3d.com
EMC uses proprietary firmware enhancements to RAID 0 (striping), RAID 1 (mirroring) and RAID S (parity protection) with field-proven scalability and performance in mission-critical environments -- 70+ of the Fortune 500. Furthermore, those RAID arrays fit logically into a seamless architecture with a robust storage software management layer with open APIs (application programming interfaces) and an effective array of services that include remote monitoring of storage systems.
The faster throughput of optical networks will also require storage subsystems to scale with it as it comes in necessary stages so as not to disrupt operations with increasingly more JIT (just-in-time) elements. That is why the deals with Lucent and Cisco are significant because EMC's technology will continue to be designed-in not only for the next generation networks -- whose architecture is a remarkably open-ended endeavor at this point since nobody expects some key enabling technologies to be available before 2005, at the earliest -- but more importantly from a revenue point of view, because EMC's dominating position will continue as the networks with legacy issues take the important transitional steps towards that eventual consensus design.
The common estimate used is that 60-70% of all carriers have significant legacy issues. New carriers like Global Crossing, Metromedia and Qwest are examples of the new breed which do not have legacy issues and are therefore aggressively pursuing new architectures. The great equalizer is that they do not have the cash flow of the branded carriers.
The evolution of the global communications backbone is important because corporations depend on it heavily to manage the flow of information. Keep in mind that these are among the most successful corporation of the world. Logistics is one key to that success -- the right information at the right time animating the right sequence of events that impact sales or costs. ATM, Fibre Channel and Gigabit Ethernet will have roles to fulfill in that kind of transition.
The darwinian processes that govern the competitive positioning of companies invariably get obscured by overly rigid and formulaic interpretations of popular titles like "The Gorilla Game" or "The Innovator's Dilemma," two well-thought out books that attempt to rationalize those darwinian processes. A complicating factor is that 1999 was a manic year for technology stocks that validated a lot of ideas, sound or not. Last year was the perfect example of how the market can be very inefficient in the short-term which in no way alters the conventional wisdom that it is very efficient in the long term.
This notion that cheaper is a necessary attribute of a discontinuous innovation, for example, is easily shredded by the fact that downtime is already approaching $600,000 per hour so any savings on the cost of the building blocks
quickly evaporates when a solution simply doesn't scale.
One must be careful to apply rigidly this notion that the same dynamics fueling the ascendancy of the cheaper Wintel platform with increasingly more powerful clustering technology in the enterprise segment will be mirrored by storage systems as they increase in size and complexity. The easy symmetry that is easy to find in the low-end and even some segments of the middle-market must necessarily be complimented by a higher-order asymmetry -- requiring field proven system architecture -- in the middle-market and high-end.
Lastly, while seemingly tangential, disk drives are the building blocks of all those RAID arrays. The chronic unprofitability of that industry has compelled the disk drive companies to develop new markets in consumer electronics and the RAID market to improve its access to the debt/equity markets and to make its equity more attractive to employees.
Seagate is the dominant enterprise vendor and provides EMC with about 90% of its high performance drives. EMC's current business plan can accomodate the quest for profitability by Seagate and the other disk drive suppliers. It remains to be seen if the server-centric vendors and the other independent storage vendors like NTAP and MTIC can do the same. The way these dynamics will play out against the backdrop of the larger competition between the server-centric IT model and the storage-centric IT model will probably determine the way that the growing number of storage networking companies -- more than 80 companies to be joined soon by McData -- will consolidate.
Dell, the quintessential box maker of the 90's, may very well be the perfect prism to watch all these sort out in free market fashion. While it currently resells NTAP's storage appliances, it is also actively looking to develop its own storage technology that will allow it to compete at the core of the enterprise. It's status as the dominant PC vendor also means that it is one of the biggest buyers of desktop and laptop drives which, in turn, means that it has to respond constructively to the disk drive industry's structural changes.
As has been discussed here many times before in the last few years, NTAP is a well-managed company that is currently executing its business plan well. To go from there to extravagant formulations of discontinuous innovations, however, is hyperbole easily dispelled by the spectrum of the requirements of the installed base and the way that the Intelligent Network of the future is being developed. |
