Call it the sign of the times, but the introduction of new specialized mainframe bundles marks the end of the platform's role as the general-purpose workhorse of the data center.
IBM this week launched a series of packages for the System z mainframe aimed at particular workloads, such as data warehousing and disaster recovery. Most of these are targeted at deploying and migrating Unix and Linux workloads at lower costs than have been available on mainframes so far. That's why the company came up with the System z Solutions Edition for Application Development, which can tap into the platforms distribution capabilities, and the System z Solutions Edition for Geographically Dispersed Parallel Sysplex (GDPS), which aims for load balancing and failover across multiple locations.
The stated reason for the move is to add new flexibility to the platform so enterprises could expand their use into more areas. However, it's hard to overlook the fact that System z sales have been on the rocks for quite a while as enterprises look to cut hardware costs with bundled commodity servers.
As the only mainframe provider in town, however, IBM shows no signs of letting the technology go quietly. The new bundles, in fact, are part of a concerted effort to align the System z with the changing nature of data center workloads. Part of that effort includes the Destination z co-development program, which recently saw the release of a new line of integration solutions by Nastel Technologies under its AutoPilot discovery and performance monitoring system. In a nod to reality, however, the release is aimed at both mainframe and distributed architectures.
When it comes to preserving the mainframe legacy, there are two important factors to consider. First, mainframes come with higher up-front and ongoing costs than commodity servers, and second, nearly half of the mainframe community runs Linux. That's why IBM and Novell are looking to cut prices on System z/SLES 11 configurations, with basic support packages cut some 40 percent to $7,200. They're also throwing in such goodies as cross-architectural debugging, disk subsystem analysis and other tools. Of course, you have to weigh this against support costs on an x64 machine, which can be as low as $349 regardless of how many cores are in place.
If you ask IBM why enterprises should continue to invest in mainframes, the stock answer will be that they offer unique capabilities and applications that can't be duplicated on a distributed platforms. Fair enough. But the fact is that for most general-purpose business applications, distributed architectures are both cheaper and more scalable, allowing you to better match you capabilities to your requirements.
And the fact is that, special programs aside, as the number of mainframe shops continues to shrink, the cost to buy and maintain big iron will continue to rise.