IBM has announced a new generation of its POWER processors, news that has sparked a breathless report from Clabby Analytics about the future of the competing Itanium chips which power HP's Integrity processors. Although HP has been enjoying a lift in its HP-UX customer base at the expense of Sun's market share, Joe Clabby's report suggests that Itanium could see even fewer iterations from Intel and HP.
Currently, HP’s operating environments are all tied to Itanium architecture. If Itanium becomes a low-volume, specialty processor, then HP’s operating environments (more specifically, HP-UX, NonStop, and OpenVMS) ultimately become low volume specialty solutions.
What will spur this low-volume status? Clabby, whose mantra has been to tout POWER and IBM since a change of heart over Itanium, thinks Intel's Xeon models will push Itanium deeper into a niche. Consider this while investing in Itanium-based Unix servers: Blades could be the refuge for the only architecture that HP-UX can call home.
Clabby has his biases, as every analyst does. He believes that the technological advantages of POWER, coupled with IBM's ardor for architecture it controls, lift POWER into the realm of Xeon acceptance. Just because it's an IBM solution, he surmises, POWER will outlast Itanium. His report from last month speculates this much.
The overlap of Xeon multi-cores with Itanium could prove catastrophic for Itanium in the general purpose processor market because Xeon multi-cores are expected to become the “industry standard” for 64-bit computing ─ relegating Itanium to the class of “special function processors” (such as IBM’s zIIP, zAAP, and IFL). These specialty processors are designed and optimized to perform certain tasks such as Java or database processing well ─ while more mainstream processors such as x86, POWER, and z processors are designed and optimized for general workload processing.
IBM says as little about POWER acceptance as HP does about Integrity business. But IBM is in control of its Identity Solution, as I like to call any technology designed and sold through one vendor. HP must accept whatever business decision Intel makes about the cost of improving Itanium, since Hewlett-Packard must pay for processors built by Intel.
Clabby believes HP is invested in design costs for Itanium's future chips, and he's probably correct. The list of vendors supporting Itanium is short about HP's purchases. Intel is likely to be requiring HP to help fund any advances its HP-UX, VMS and NonStop customers require. Clabby even has a scenario for how HP will manage the expense of maintaining the Itanium line.
To reduce design costs, HP may ultimately be forced to use its blade chassis as the only system design for
Itanium-based servers. Releasing Itanium-based servers in a blade form factor would make a lot of sense
because HP is already investing heavily in blade design to support its x86-based server line ─ so adding
Itanium-based specialty processors to existing blade chassis would minimize HP’s Itanium-based
system/server design costs.
The bottom line is to insist on a blade implementation for any Integrity server you'd buy in a migration. It might seem like a no-brainer solution. But hardware costs can be an element that a packaged solution supplier would try to reduce, selling a non-blade Integrity. HP has five models of Integrity systems it sells for HP-UX that aren't in a blade form factor. The vendor has a concise Integrity Systems Family Guide which explains the range of this HP Identity Solution. POWER7 may help tip the scales of market share more toward IBM. HP's blade commitments will make its bladed chassis the most weighty part of Integrity's counterbalance.