EMC's price premiums are easily misunderstood by people who narrowly focus on just the hardware or software "that only runs on Symmetrix." EMC derives its price premiums from a combination of hardware, software and services that are seamlessly integrated to provide what it calls an infostructure, an system architecture field-tested over the years in 25,000 customer sites and which provides the following benefits:
1) Maintain and preserve the customer's option of plugging into the most powerful computing platform available.
The mainframe market is slowly declining from a current installed base of about 60,000 units, but the parallelism-oriented competition between IBM, Hitachi and Fujitsu continue at a heated rate with MIPs declining at an even faster rate than Moore's Law. Over the years, the IBM-mainframe compatible knowledge base and talent pool have been aggregated by the few remaining mainframe players and EMC, the only independent storage vendor still catering to this market. This type of scarcity partly accounts for the price premiums of Symmetrix.
2) Maintain and preserve the customer's ability to plug into the UNIX platform.
3) Maintain and preserve the customer's ability to absorb the superior economics of Wintel and to a certain extent, Intel+Linux.
4) Allow the customer to change and fine-tune the combination of computing platforms depending on price and performance.
The dynamic range of rigorously field-tested features, particularly software, allows EMC the unparalleled ability to customize its solutions and sustain its premiums because Symmetrix resides at the core of the customer's networks.
After EMC surpassed IBM to become the preferred vendor to the IBM-mainframe compatible market (IBM, Hitachi, Fujitsu, etc), it then proceeded to become the first and preferred storage vendor to the heteregeneous server market (IBM-mainframe compatible + Unix + NT.....). EMC's overall technology lead over the rest of the competition is widely considered to be two years, but the software is now at an annual run rate of over $1 billion a year and that is expected to, at the very least, preserve its lead while services can only add to it.
The author shows his misunderstanding of the EMC family of solution when he lumps EMC with other "directly-attached server/storage" vendors when the fact of the matter is that EMC has always provided consolidated storage that supports 35 operating systems while its closest competitor supports just 5. Of course, if one turns logic on its head and try to narrowly assess the very popular and broad Symmetrix product solutions solely by its ability to hang off the TCP/IP network transport layer then it's easy to lapse into that conclusion even though the very existence of the Infineband project clearly demonstrates that TCP/IP is not mature yet to meet the stringent requirements of the high performance enterprise storage market. Infineband is 3-5 years away and even then that type of industry standard has to be rigorously field-tested and engineered to meet the requirements of the installed base at that point in time. That is always a tricky affair.
NTAP is a well-managed company with an innovative thin server solution with proprietary and tightly coded software that has met the requirements of its target markets (arrays with 2-60 disk drives) so far, but the nature of the network is changing. Bandwidth, for instance, is becoming cheaper and more abundant. Applications and processing power are already being readied to soak that bandwidth.
Clearly, storage growth is going to be exponential post-Y2K and that changes the rules of storage at the low-end, the mid-range and the high-end in EMC's favor.
Clearly, NTAP doesn't have a "radically different set of offerings altogether." To accept that statement at face value would mean that the high-performance storage market has never encountered the problems that NTAP ostensibly solves. How reasonable is that?
Take these statements for example, and compare the features with EMC's solutions.
For example, Data ONTAP has integrated within itself a patented feature known as Snapshot, which automatically creates one or more read-only file backups within a given file server, backups that take only a fraction of the disk space of the original.....
.....Known as clustered failover, it involves using two Netapp filers which on a day-to-day basis act completely independent of each other. Each has its own processor, operating system, and hard disks to manage, and data requests made to one filer consume none of the resources of the other.
Timefinder is 4th generation software designed for the mainframe and mixed-server market. It clearly provides more functionality and on a much larger scale. Data ONTAP and clustered failover may be proprietary but every high-performance storage vendor has had proprietary variations of those capabilities for some time.
TimeFinder is unique software from EMC© Corporation that supports mainframe, open systems, Windows NT©, and AS/400© platforms. It allows system and storage administrators to create, in background mode, independently addressable Business Continuance Volumes (BCVs) for mainframe and open system information storage. The BCVs are mirror images of active production volumes that can be used to run simultaneous tasks in parallel with one another. This parallel processing capability offers workload compression and allows you to significantly increase your company's efficiency and productivity while maintaining continuous support for the needs of the enterprise.
Once BCVs have been created, they can be split from their production mirrored volumes and used for such tasks as loading data warehouses, testing new applications, running batch jobs, backups, and other functions that typically cause a disruption to normal business support. They can also be used to restore the environment following testing or system changes. BCVs can be locally mirrored (RAID 1) for additional protection.
EMC TimeFinder keeps your business running at full capacity even while scheduled outages occur. Once the task on the BCV is complete, the volume can be resynchronized with the production volume, reassigned to another production volume, or maintained "as is" for another task.
emc.com
Navisphere Application Transparent Failover (ATF) software, when running on a server connected to one or more EMC CLARiiON disk arrays, delivers automatic re-routing of I/O traffic in the event of a host-to-storage path failure. Failover to a redundant I/O path is transparent to the server's application software and users. With ATF, your business operations are not impacted by the failure of a cable, host bus adapter, or storage processor. Instead, data continues to be delivered without interruption. Failover occurs in seconds, permitting continuous access to the information stored on the disk array(s). Navisphere ATF is an ideal complement to host clustering packages such as Microsoft Cluster Server, IBM's HACMP, Sun Clusters running Veritas FirstWatch, and HP's MC/ServiceGuard.
And, EMC CLARiiON's patented design provides two additional levels of data integrity beyond other RAID products: end-to-end checking of the data inside the disk array and parity coherence. End-to-end checksum provides a mechanism against data corruption, even after unexpected events occur. Data parity coherence protects against the possible inaccuracies that can be created during power outages and disk failures.
emc.com
If price were the only metric used by the Global 2000 then clearly EMC wouldn't be where it is now because Symmetrix is premium-priced.
If modularity were the only metric used by the Global 2000 then clearly EMC wouldn't be where it is now because it provides a seamless architecture that combines hardware, software and services customized specifically for each customer.
Ease of use and speed are relative terms depending, for example, on scale: file sizes, the size of databases and the unique usage patterns of each organization. What has consistently been proven over the years is that corporations go through cycles of allowing computing devices to proliferate and then consolidate as more sophisticated hardware and software automation becomes available. This has been proven true in tape devices, optical devices and even NT servers. DG's Avion, for example, is being streamlined to provide high-powered Wintel servers using NUMA shared memory technology to consolidate older Wintel servers.
However, the tide does appear to be turning in the company's favor. Their annual sales growth recently accelerated into the triple digits, way ahead of the 30% growth the storage industry's experiencing
The reference to the turning of the proverbial tide and the use of percentages apart from the small numerical base just about sum up the basic flaws of comparing EMC's architecture and NTAP's modular approach as if it were mutually exclusive approaches in a storage market that has clearly been moving in EMC's direction for a long time. The novelty of what the author considers NTAP's key proprietary features has clearly been overstated. In doing so, he misses the real war intensifying betweeen EMC's storage-centric architecture and the OEMs' one-stop shopping server-centric architecture -- financing, hardware, software and services.
Instead of prematurely pitting EMC against NTAP at this early stage of the storage boom, this author would do well to focus on the truly disruptive and vise-like threats to NTAP posed by the one-stop server OEMS AND by NTAP's own disk drive suppliers.
The drive makers are a particularly dicey proposition for NTAP because they have basically claimed the low-end of the SAN/NAS market as their own due to their sheer lack of profitability in their traditional businesses and the higher margins (2-3x) and growth rates (50+%) of the low-end SAN/NAS market. Keep in mind that storage is expected to eventually account for 75% of the annual IT capital expenditure.
While their products may not yet provide the same features as NTAP, they have clearly signalled that they're willing to buy more start-ups to bolster their offerings. No drives, no arrays. What can I say?
EMC and the other server OEMS do not have the same problem because of the sheer size of their annual purchase orders; although, some of the server OEMs may grumble at the way their bloodied disk drive vendors are moving in. But what can they do? Their drive guys can't make money on any sustainable basis and that eventually affects R&D.
At a revenue level of $4 billion, EMC, for instance, had a total bill of materials of about $1.6 billion, or 40% of revenues. On its way to revenues of $12 billion by 2001 and with EMC Software approaching $2 billion in revenues, one can reasonably expect EMC's total bill of materials to increase to around $4 billion. That's a lot of buying power and clout enhanced by EMC's 2-year technology lead. IBM, for example, recently disclosed that it lost $350 million worth of business because it couldn't ramp up its 10k rpm drives fast enough to demand. In contrast, EMC doesn't shop its disk drive business around and has long maintained a tight workinging relationship with Seagate that allows them priority access to SEG's high-performance drives. |
