Through the first half of this year, the server market was nothing short of a disaster.
Here’s IDC’s firsthand account of the carnage:
According to IDC’s Worldwide Quarterly Server Tracker, factory revenue in the worldwide server market declined 30.1% year over year to $9.8 billion in the second quarter of 2009 (2Q09). This is the fourth consecutive quarter of revenue decline and the lowest quarterly server revenue since IDC began tracking the server market on a quarterly basis in 1996. Server unit shipments declined 30.4% year over year in 2Q09, accelerating from the 26.5% decline experienced in 1Q09 and representing the largest ever year-over-year quarterly server unit decline as customers continued to defer server refresh activities.
It was bad news for all the vendors, but particularly ugly for Sun Microsystems. As InfoWorld reports, Sun suffered more during the quarter than any other top-five server vendor. Its worldwide server revenue fell 37.2 percent, and the slide is sure to continue as long as uncertainty remains about the ultimate disposition of Sun’s hardware business resulting from the Oracle acquisition. As noted previously on this blog, HP and IBM have been passionately seducing Sun’s increasingly anxious customers.
Sun’s company-specific problems notwithstanding, we should be under no illusions: The server market was not kind to anybody in the second quarter. Losses occurred everywhere, regardless of vendor, of class of system (x86 or non-x86), of server size, of target market, or of geographic territory. It was an all-out bloodbath.
Signs of hope? Thankfully, one or two exist.
Blade servers aren’t doing great, but they did better than other server variants. Shipments of blade servers decreased 19.8 percent year over year; revenues dropped 12.1 percent, representing the second consecutive quarter in which the numbers have gone the wrong way. Still, it was a relative bright spot.
IBM is said to have gained 3.8 points of market share in the blade segment during the quarter, but HP remains the dominant blade player, holding 52.9 percent of sales compared to IBM’s 27.2 percent and Dell’s 9.1 percent.
In the server market as a whole, IBM retained the pole position with 34.5-percent of factory revenue. HP maintained second place with 28.5-percent share for the quarter; it suffered a 30.4-percent year-over-year revenue decline. Dell and Sun held the third and fourth positions, with 12.4-percent and 10-percent shares of factory revenue, respectively. Dell’s revenue declined 26.8% and its market share increased marginally. Sun, as noted above, lost more ground than did any other major vendor.
IDC sees reason for optimism in the ancient server installed base. It suggests corporate IT departments can only hold together these museum pieces so long before they will have to replace them. That bodes well for a refresh cycle.
Then again, when that cycle materializes, it will not be like those of the past. We are unlikely to see one-to-one server replacement rates. Instead, with blades and virtualization clearly ascendant, vendor pickings will be slimmer than in past refresh cycles. Vendors with the strongest virtualization pitches, including the right mix of enabling software and services, will be well placed to take home the business.
Some wonder why Cisco is getting into this consolidating, cutthroat, Dickensian world. The move makes a certain sense, though, and not just because Cisco is seeing diminishing returns in its enterprise-networking principality.
What’s happening in the server space — blades, virtualization, cloud computing, an emphasis on data-center consolidation driven as much by long-term economic reality as by technological advances — is genuinely disruptive (a word that is abused and misused extensively).
An opportunity exists for a vendor to pull together all the data-center pieces and fuse them into an easy-to-use, cost-effective whole. That’s the business rationale behind Cisco’s Unified Computing System (UCS). But, in the end, it’s probably driven as much by fear as by greed.