The end of the mainframe has been predicted too many times to count over the years, so it's with a jaundiced eye that I view any pronouncement of the workhorse's sudden demise now.
Still, it's fair to say that mainframe use will probably continue to dwindle in the coming year and that the tasks and functionality of those that remain won't be the same as the storied systems of the past.
As with everything else in the data center, the cloud will likely be a major driver in these changes. Information Week's Roger Smith probably overstates it a bit when he says Micro Focus' decision to release cloud-friendly Cobol apps will lead to a wholesale "unplugging" of the mainframe. But the utility-style benefits that the cloud brings will most certainly cause many enterprise managers to rethink the types of apps and data that are on their big iron. It's a safe bet that anything non-critical is fair game for the cloud.
Any trends targeting the mainframe will certainly be welcome by those who have a vested interest -- namely HP -- in seeing it go the way of the dinosaurs. The company has been talking up its "mainframe alternative" strategy quite a bit lately, telling audiences that the only way to establish the kind of flexible and dynamic IT needed for 21st Century business environments is to ditch proprietary technologies and embrace net-based, platform-standard technologies.
Not everyone is buying that argument, however. Korea Exchange Bank recently powered up a new System z10 for its credit card transaction infrastructure. The company said it was drawn to the triple-unit system due to the fact that it provides the equivalent of nearly 1,500 x86 servers at only a fraction of the floor space and power consumption. In addition, IBM was able to migrate the bank from the previous z990 system without interrupting bank operations.
The one gripe that continues to dog the mainframe, however, is poor management capabilities, particularly when it comes to controlling the rate that related systems consume processing power. Without a better way to manage data, users are forced into a steady stream of CPU upgrades to keep up with demand.
IBM has kept up with regular improvements to its Tivoli management stack, driven largely by new-found competition from third-party developers like CA and BMC. CA recently upgraded its NetMaster toolset to address some of the deficiencies that IBM systems have in dealing with SNA transport protocols needed for monitoring legacy apps on IP networks.
Is the mainframe ready to retire? Certainly not. Will its role in the datacenter change? Surely, but that's been a fact of life for the mainframe for close to 40 years now.