If anyone thought mainframes were passé, news from IBM last month set the record straight. And it’s prompted questions from a few enterprise IT pros as to whether a mainframe is a more reasonable approach than a private cloud. After all, mainframes run 70 percent of the world’s mission-critical business applications today, so why shift to another approach?
First, consider IBM’s announcement on August 28: The zEnterprise EC12 mainframe, which features the world’s fastest chip (5.5 GHz), can run hybrid workloads of real-time transactions and near real-time analytics. These capabilities further augment the mainframe’s present ability to run a hybrid of both mainframe operating systems and distributed operating systems like Windows, Linux, and Unix -- all within the mainframe footprint, via an integrated zBX Bladecenter.
Does it sound a lot like cloud in a box? It’s not far from it.
Yet, the news still has the ability to startle a generation of IT’ers who have cut their teeth in a distributed computing environment of Intel servers running Linux, Unix, and Windows -- which they see as the pathway to the cloud. Why would you even use a mainframe if you have servers to run the cloud?
The best way to answer the question is to review the main function of servers, i.e., to handle requests for computing resources and data. In this sense, mainframes qualify as servers as much as any other type of computer. Add to this the fact that virtualization, a cornerstone of cloud computing, originated and existed on mainframes as early as the 1960s.
Regardless of this mainframe heritage, most cloud infrastructures, even in mainframe shops, are being founded on distributed computing platforms first. Not every IT department has a mainframe, and for those that do, it's taking additional time to determine exactly how they want to plug their mainframes into their private clouds.
Still, there are enterprises and even cloud providers basing their networks on mainframes. For example, Oildex subsidiary Transzap, a cloud-based SaaS (software as a service) provider, has opted to use a mainframe in its data center to run virtualized Linux -- even though the company had no prior history of running any mainframe systems. The cloud provider likes the mainframe’s reliability, capability, and overall lower cost of ownership.
Many new banks and telecom companies in developing countries are selecting mainframes because these mega-servers are still best in class for the mission-critical applications that these organizations run.
Mainframe services are also popping up in the cloud; a range of SaaS providers provision mainframe resources on an on-demand basis. First National Technology Solutions, for instance, provides managed hosting of mainframe resources to companies that have their own mainframes, but require additional temporary mainframe resources for activities like application testing.
There are plenty of specific business cases that make an argument for the mainframe as a server in the cloud.
Does this mean that mainframes will challenge other servers in cloud presence? Hardly. Even IBM clearly indicates the opposite, as its PureSystems solution released earlier this year indicates. The culmination of a three-year, $2 billion research investment, PureSystems is based on an Intel server foundation and is being touted as a pre-configured and pre-optimized “Smart Cloud” solution for enterprises and SMBs (small and medium-sized businesses).
The bottom line here, for both enterprises and cloud service providers, is that there are many different virtual and physical server choices for private cloud development. Most of these server choices will come from Intel-class machines. But for many others, the mainframe will be a cloud-player fully capable of “serving it up” as well as any other server out there.
— Mary E. Shacklett, President, Transworld Data