As solid as HDS is at the high end of the storage market, it is missing some key pieces, the biggest of which is homegrown storage management software. HDS and its parent company are currently negotiating to buy such a company, Roberson says.
HDS is probably the most likely buyer for DataCore, IMO.
Here's a little fable on storage software acquisitions.
StorageTek and Encore Computer started working on making storage systems that would support both mainframe and open systems in 1990. Amdahl joined the effort a few years later as EMC came out of nowhere to grab the lead in mainframe storage, from less than 1% of an $8B mainframe storage market in 1990 to 41% of a $4B mainframe storage market in 1995.
By 1995 much of the development work on the data-sharing consortium had been completed and the collaborative result, the Iceberg, was being extensively tested by StorageTek.
By 1995 also, IBM's market share of a shrinking mainframe storage market had collapsed from around 75% in 1990 to 31% in 1995. An IDC report at that time pegged EMC as the performance leader with IBM and Hitachi tied for leadership in terms of reliability. As a result, IBM decided to launch a massive counteroffensive against EMC, but IBM couldn't scale its RAMAC controller technology to keep up with Symmetrix so IBM did the unthinkable and entered an alliance with StorageTek 1997 to resell the Iceberg, which by then had been delayed by about 2 years.
That 2 year delay was more than enough for EMC to come out of nowhere, AGAIN, to grab the lead in open systems storage. EMC's market share went from 0% in 1994 to 16% in 1996 and it has been gaining market share every year since then.
As many here know, EMC entered the open systems market with alliances with NCR and HWP. HWP was then the dominant Unix vendor; although, Sun was in the process of exploiting the way HWP, IBM and other Unix vendors were starting to underinvest in Unix in anticipation of Itanium (formerly Merced due in 1999). As I recall, in terms of units, Wintel servers nosed ahead of Unix servers in 1996; although, in terms of revenues, Unix servers still had a commanding 6 to 1 advantage.
While StorageTek was testing Iceberg, Encore had decided to put itself up for sale in 1995 but found no interest from its former parters, StorageTek and Amdahl. Sun ended up buying Encore for around $200-300M in 1997 and Encore's data-sharing technology became the foundation for the ill-fated A7000 which, as I recall, had more than 100 beta sites up and running before Sun finally pulled the plug in 2000.
Despite close to 10 years in the hands of the best engineers from Storagetek, Amdahl and Sun, Encore's technology proved too difficult to integrate with Solaris. So Sun, which had acquired Maxstrat in early 1999, started working on Plan B - Purple.
The StorageTek and IBM reseller agreement ran for 2 years. By 1999, IBM rushed Shark to market even though it did not yet have the complete feature set to compete with EMC.
Curiously, before IBM could even deliver on the Shark features that it promised over an 18-month period, it had already decided to talk up their Storage Tank virtualization technology, again due over an 18-month period. WitSoundview estimated that Shark revenue (hardware and software) was around $300M in IBM's most recent quarter. This a remarkably anemic run rate considering the persistent field reports and statements from IBM that they were pricing the toothless Shark as much as 90% below Symmetrix!
The point of this walk through the recent past is to point out that storage acquisitions based on technology are very difficult endeavors and more often fail than not.
Hitachi seems determined to compete directly against EMC which it publicly declared as its main archrival earlier this year. I think that's a mistake because Hitachi is not only competing against EMC and its $1B laser-focused R&D budget, it is competing against EMC's partners in its Open API program which will produce twice the number of products next year (70+) that it produced this year (30+).
Fujitsu's approach is probably more interesting. Fujitsu's storage software subsidiary -- Fujitsu Softek -- is expected to gross $90M in sales in its full year of operations and there are indications that Fujitsu may join EMC's open API progam rather than compete head-on. EMC and Fujitsu already have mutliple alliances in different parts of the world. One reason I think that this is the smarter move is that one uncharted area for storage software is wireless data. NTT Docomo is leading the charge to 3G with over 22 million subscribers. |
