The question of whether it makes financial sense to migrate existing 32-bit system architecture to newer 64-bit offerings depends on the operating system and application support, according to Graeme Dendy, product manager for enterprise servers at Business Connexion`s Persetel Competency.
"The motivation for introducing 64-bit processor systems into a business would be to improve the throughput from applications by increasing the amount of memory that can be addressed per transaction. However, this level of throughput is only ever really demanded in the high-end computing space where 64-bit in the Unix arena has been prevalent for many years," says Dendy.
Dendy maintains that demand in the mid- to low-end arena will be driven as and when applications and the Windows operating system migrate to a 64-bit platform.
"Microsoft already has a 64-bit version of Windows 2003 Server available to run on Intel`s Itanium processors; however, the 64-bit Windows/Intel (Wintel) market is not taking as much of the Unix/Risc share as was expected," he comments.
He says this slow uptake in the Wintel space can be attributed to the cost and support issues. "For the cost of migrating to an all-64-bit architecture, the difference between Risc and Intel in terms of return on investment (ROI) and total cost of ownership (TCO) are not justified as yet," he says.
"Add to that," he says, "the limited number of 64-bit Wintel applications or the cost of porting applications to a 64-bit platform, and the market has seen little or no need to make a sudden shift to 64-bit architecture."
That is not to say that 64-bit Wintel architecture does not belong in business. Dendy says 64-bit is inevitable in the not-too-distant future so it is just a matter of time before all operating systems and applications are ported, according to market demands.
"Traditionally, a server should have a life span of about three to five years. Given the current slump in the ICT market, where CIOs are mandated to sweat assets and derive as much value from existing investments as possible, the best approach for implementing 64-bit in the business would be to first assess the need and support from a software perspective," Dendy advises.
"Once the business needs for 64-bit have been identified, 32-bit servers should be phased out and gradually replaced with 64-bit servers when their lifecycles end," he adds.
A similar argument could be presented for the migration to 64-bit on the desktop.
The lack of 64-bit requirements from desktop operating systems and applications is not creating a need to bring this technology way down to the desktop as yet.
AMD, however, is making a lot of noise around its new Opteron and Athlon 64 processors, which, it says, are fully backward compatible to 32-bit applications. This means that users can still have the full functionality of their 32-bit applications and will have made provision for the 64-bit future.
"One element that may push 64-bit to the desktop and mid-tier server could be the increased need for memory addressability for graphics and video," Dendy says.
"If money was not an issue and the business demanded increased throughput performance was a requirement, a businesses should look at investing in 64-bit now rather than later. However, in the real world, a phased approach, taking into account operating system and application support on all levels, is the best," Dendy concludes.
Share
Editorial contacts