Although the Alpha technology was good, Compaq Computer has backed out of its costly production, as Gartner predicted. Compaq will move its entire operating system (OS) family to the Itanium chip and send the Alpha engineering team to Intel. Compaq needed to make this move, but Intel will benefit most.
On 25 June 2001, Compaq and Intel announced an agreement that will move Compaq’s entire OS family to Intel’s Itanium chip by 2004. Intel will offer jobs to most of Compaq's Alpha staff. Compaq will start work immediately to port Tru64, OpenVMS and NonStop Kernel Himalaya operating environments to Itanium.
This move stems from Compaq’s analysis of the long-term Intel 64-bit processor road map, coupled with its desire to cut costs. The company faced an uphill battle generating enthusiasm among independent software vendors (ISVs) and difficulty marketing a diverse set of OSs and hardware platforms. The announcement boosts Itanium's attractiveness among ISVs and original equipment manufacturers (OEMs). Although the move may allow Sun Microsystems to skim Compaq's business in the short term, in the long run, it will increase pressure on Sun by letting Intel gain credibility as a supplier of enterprise platforms.
Intel will initially use the hundreds of engineers acquired from Compaq to improve compiler design and software development tools for the Itanium Processor Family (IPF). Over time, this talent will also contribute to future IPF processor designs; however, lockdowns in the McKinley, Madison and Deerfield processor designs mean the additional brainpower will have no significant impact until at least 2004.
To stem potential customer defections, Compaq will need to port OpenVMS, Tru64 Unix, and the related clustering and system software (e.g., workload management) to Itanium as soon as possible. Customers and ISVs will have to recompile Tru64 applications; OpenVMS applications can be recompiled or run through a binary code translator. Unlike Hewlett-Packard (HP), Compaq does not plan on providing a binary code translator for Tru64.
Although this agreement resolves the nagging issue of Alpha's viability, Compaq is not out of the woods. As far back as 1998 Gartner warned that Compaq's commitment to Alpha would safely hold up only until 2003, with follow-on technical support for five to seven more years (see Research Note SPA-05-4366 “The Alpha Roadmap: Secure for Now, But Is it Safe?”). Gartner recommended making only tactical commitments to the platform for a three to five year period and said that Compaq's management might change course and back Intel across all OSs. Now that this has happened, Gartner offers the following advice:
Enterprises should upgrade Tru64 Alpha servers as needed, minimize new deployments and consider new applications only when the ISVs provide a written commitment to transition the application to the Itanium environment. Enterprises using Tru64 and those in the midst of Alpha deployments should seek ISV and technical support similar to Oracle's commitment to support both Alpha and Intel concurrently when the first generation Compaq Tru64 Itanium systems appear. The full extent of Compaq’s commitment to facilitate transition will be unveiled in July and August 2001; enterprises should take this into account as they re-evaluate their commitment to remain on Tru64.
OpenVMS users will have the most difficult time moving applications and middleware to Itanium. Compaq must alleviate concerns that cluster support and the large base of system software will be available for Intel before Alpha delivery stops. Compaq plans to address these concerns by guaranteeing users that required Alpha-based software will run in the Itanium environment. Enterprises should factor the scope of these guarantees into their evaluation to stay on OpenVMS or transition to a new environment.
NonStop Himalaya may face an easier migration due to prior work Compaq performed to port NonStop to Intel processors; however, Compaq has some unfinished work to complete on application-specific integrated circuits to accommodate the fault-tolerant architecture design.
The Intel deal clears the air about Compaq's server platform stability and direction, but Compaq still faces problems of Tru64 acceptance among users and ISVs. More importantly, Compaq must quickly develop and implement an effective transition program and maintain the interest of the installed base in Alpha until cutover. Failure to execute could hasten a sharp decline in Unix server revenue.
Intel emerges a big winner here in the battle for “mind share” among OEMs and ISVs. Compaq can now declare that it has "all the wood behind the (right) arrow," leaving Sun as the only vendor lacking an Intel option except in appliances. HP, whose work with Intel continues regardless of the deal, benefits from the advances derived from the influx of Alpha talent but has a head start in the transition process. IBM remains the only contender that can challenge Intel on the microprocessor battlefield.
Analytical Sources : George Weiss, Unix & Midrange Strategies, and John Enck, NT Strategies