Sales of Hewlett-Packard’s business critical systems (BCSs) powered by Intel Corp.’s Itanium central processing unit hit their six-year low in the second quarter of HP’s fiscal year. In addition to global economy slowdown, the reason for the decline of sales is conditioned by customers’ reluctance to buy systems based on IA64 architecture, which many believe is approaching its end-of-life.
“Business critical systems revenue declined 37% year-over-year, driven by continuing secular pressures and the weakness in Itanium servers,” said Catherine Lesjak, chief financial officer of HP, during the most recent conference call with financial analysts.
During the quarter, which ended on April 30, 2013, sales of mission-critical HP Integrity servers and Superdome machines declined to $266 million, the lowest since fiscal year 2007. In the previous quarter HP’s BCS division’s revenue was $306 million.
Shipments of HP’s Itanium-based systems started to drop in 2011, when Oracle announced intention to stop making software for IA64 servers citing HP’s and Intel plans to stop development of Itanium products in favour of new x86-based chips with mission-critical enhancements.
Despite of the fact that the court have ordered Oracle, one of the world’s top supplier of enterprise software, to continue making programs for Hewlett-Packards business critical systems based on Intel Corp.’s Itanium chips, demand towards such servers is not increasing. Since Intel started to add mission-critical features to Xeon-class microprocessors, it is evident that at least certain customers preferred Xeon-powered business-critical machines. Other customers decided to stick to “wait and see strategy” and invest into neither of the platforms.
Earlier this year Intel quietly announced it would delay development of long-expected unified platform for mission-critical computing that would be compatible with both Itanium and Xeon central processing units (CPUs). As a result, the next-gen Itanium code-named Kittson will be made using current-gen process technology and will be compatible with existing Itanium platforms.
The most important thing about the next-generation Itanium known as “Kittson” is that it was supposed to be made using 22nm or 14nm process technology, thus, significantly boosting clock-speed and/or core-count. Being made using 32nm fabrication process will likely result in moderate increase of clock-speed and hardly into core-count boost. Even though Kittson in its current form will be compatible with Itanium 9500 “Poulson” sockets and thus will provide upgrade path for existing systems, it is unlikely that Itanium customers will be glad with 20% performance increase, given the rapidly growing need for performance.
The plan to make Itanium and Xeon socket compatible has been Intel’s plan since early 2000s and in the recent years the intention became obvious and natural as Intel started to incorporate technologies for mission-critical computing into Xeon enterprise/expandable platforms. Unification might brought Intel and its partners a lot of benefits and greatly simplify development of server platforms and catalyze software makers to port their mission-critical software to x86 chips. On the other hand, unified platform would allow Intel to cost-efficiently develop new versions of Itanium provided that there is demand for such processors.
With ephemeral upgrade paths for current Itanium-based machines, many decision makers these days may become more interested in reducing reliance on Unix systems and deploy more machines with Intel Xeon E7-class processors that feature numerous RAS (reliability, availability, serviceability) features. Still, those, who need HP-UX, OpenVMS or NonStop machines are forced to utilize both today’s Itanium 9500 “Poulson” as well as next-generation “Kittson” products.
Nonetheless, no matter how dependent many clients are on Unix and Itanium, they clearly either do not want to purchase new systems today, or are simply migrating to other platforms from stagnating IA64.