One thing EMC has not done is fight HDS by expanding its partnerships.
Huh? First of all, the June announcement between the 4 storage vendors -- EMC, IBM, HDS and Compaq -- and the two switch vendors -- Brocade and McDATA -- covered joint support agreements, not the type of microcode sharing that already occurs regularly in the mainframe environment.
For example, the 1999 agreement between IBM and EMC was prompted by joint customers who wanted IBM's mainframe clustering technology to work better with EMC's SRDF so IBM and EMC agreed to swap APIs for those applications. IBM and Hitachi have also swapped remote copy microcodes for many years to compete with EMC's SRDF with limited success.
70% of the Symmetrix installed base is in mission-critical environments so it comes as no surprise that EMC has already developed more than 40 joint support agreements with a variety of vendors including the likes of Cisco and Oracle in EcoStructure. EMC's open API program also has more than 80 ISVs already shipping more than 40 software products with accompanying joint support programs.
Everyone’s coöperating except EMC,” Lehman’s Foster says. “And they’re going to lose. They’re losing already.”
This is nonsense. EMC started supporting mainframes, from IBM and HDS in 1990, and mainframes, Unix servers and NT servers from all vendors in 1995. IBM, HDS and Compaq only started to support each other's servers last year so to even suggest that they are now in a position to support each other's storage systems is nonsense. It also ignores the fact that EMC is already shipping its ESN Manager which already supports the leading switches and the leading arrays; albeit, at a very rudimentary level (zoning control, path control and volume access).
The fact of the matter is that if a vendor can only support a couple of server platforms then it can only support a few database platforms and only a few applications that use those databases so what practical use is the ability to make one's storage systems communicate with the storage systems of other vendors?
For example, IBM can only support its own mainframes, its own Unix servers and its own NT servers, but it has an impossible time supporting the servers of other vendors except for the servers of its partner-of-the-month.
It doesn't end there.
Since IBM is also selling its own database against Oracle, NCR, Microsoft and Sybase, it has an impossible time supporting the servers from other vendors that also run those competing databases.
It doesn't end there.
Since IBM is selling its own middleware against BEAS, Sun and Microsoft and Oracle, it has an impossible time supporting servers that run those databases and middleware.
70% of all enterprise applications use the relational database as the foundation so those competitive disadvantages from the physical storage layer to the applications layer can quickly add up.
Given all that, the ability to support other storage systems is pure marketing puffery.
Because HDS isn’t traded separately, there’s no way to know how its margins compare to EMC’s, but it’s clear that EMC’s are shrinking as a direct result of HDS’s focus on the space.
There are enough numbers out in the open to conclude that Hitachi is heavily subsidizing HDS so EMC has to employ the same tactics that IBM used to drive HDS out of the mainframe business last year.
Right now, Hitachi takes care of system level R&D while HDS takes care of storage software R&D so HDS R&D numbers and cost structure are meaningless.
Hitachi also supplies HDS with disk drives and DRAMs. Hitachi only has 3% of the overall disk drive market. The rule of thumb is that a disk drive operation only starts to become profitable at 10% market share so it's reasonable to assume that Hitachi is heavily subsidizing HDS in disk drives since I don't think HDS can wage its price war against EMC with Hitachi disk drives priced at arms lenght. By contrast, EMC gets its disk drive supply from Seagate and Fujitsu ,which control around 60% of the enterprise drive market. Although accounting for only 11% of disk drive industry unit shipments, the enterprise drive segment is the most profitable with gross margin routinely in the 20% to 30% range.
In addition to having no manufacturing scale, there are clear indications that Hitachi is also falling behind technically in terms of areal density (track density x linear density) and data transfer speed (linear density x rotational speed) so Hitachi is looking at the prospect of throwing more good money at a familiar manufacturing vicious circle -- no scale, no technology, no scale, no technology etc -- in probably the most brutal industry in IT: the disk drive industry.
Lastly, EMC booked $1.4B in software revenue last year compared to HDS' $50-$100M. Through June, EMC has already booked close to $1B in software revenue. Assuming 85% gross margin, EMC has a software margin advantage of close to 19 full percentage points against HDS so even if one assumes that EMC and HDS have comparable hardware gross margins and service gross margins, HDS is clearly at a margin disadvantage of such magnitude that, more than anything else, indicates Hitachi's commitment to this market. |
