Mainframes Still Matter in the Digital Business Transformation Age
The most venerable and arguably more conversational platform in all of enterprise IT remains the mainframe. Now well into its sixth decade of existence IBM claims mainframes still handle almost 70% of the world’s production IT workloads. Nearly three-quarters (74%) of the 200 IT and business executives in North America recently surveyed by the market research firm Oxford Economics on behalf of IBM said mainframes are still central to their IT strategy.
The bulk of the most mission-critical applications running on mainframes are online transaction processing (OLTP) applications that require a level of processing capability that still can’t be easily matched using any other platform. Many of those OLTP applications have over the years been infused with analytics capabilities that surface insights in near real time. There are plenty of instances where certain classes of workloads have been migrated off the mainframe to run on lower cost platforms. However, there are still many enterprise applications that continue to run faster on a platform that traces its lineage back to the 1960s.
A New Era for Mainframes
In the era of digital business transformation mainframes are entering a new phase. A wide range of applications running on a variety of distributed computing platforms now need to access data that resides on a mainframe. As a result, mainframe applications are being modernized using application programming interfaces (APIs) to enable developers to achieve that goal. IT teams that manage mainframes are starting to adopt best DevOps practices to increase the rate at which applications running on the platform can be updated in lock step with applications running on distributed computing platforms.
At the same time, IBM has been making a concerted effort to enable other classes of applications to run on the platform at a lower cost. IBM makes it possible to both deploy Linux in emulation mode on the z/OS operating system that drives the bulk of the applications on the platform today in addition to making available instances of the platform that only runs IBM LinuxOne, a distribution of Linux optimized for the mainframe platform.
Most recently, IBM has also embraced Docker containers to provide another means to run Linux applications on z/OS. Docker containers encapsulate applications in a way that enables them to run on any platform. IBM, with the recent release of version 2.5 of IBM z/OS, significantly improved the performance of containerized applications running on top of that operating system. That approach also eliminates the need to provision separate logical partitions (LPARs) or system images to run Linux applications in emulation on a mainframe.
Regardless of approach to enabling other classes of applications to run on the IBM platform, most of those efforts are primarily designed to entice organizations that have Java applications that typically run on a Linux platform to move them to a mainframe. The pricing structure that IBM employs for mainframes is based on peak usage of the platform. That model makes it possible to run some applications at essentially no additional cost alongside, for example, a high-performance online transaction processing application (OLTP) running natively on z/OS that doesn’t fully consume all the available capacity on the platform. That approach also reduces the overall latency of the application environment when, for example, a Java application needs to access data created by an application written in COBOL running natively on z/OS. IBM also recently added support for Java/COBOL interoperability that enables parallel 31-bit and 64-bit addressing to enable these applications to run simultaneously on the platform.
Also read: AI and Observability Platforms to Alter DevOps Economics
Gaining Developer and IT Support
Convincing developers that the mainframe is in effect just another platform they should invoke is crucial for IBM. The number of developers writing applications in COBOL is not growing at the same rate as other programming languages. In fact, a recent survey of more than 80,000 developers conducted by Stack Overflow, an online community for developers, identifies COBOL as the language developers dread the most (84%). Responsibility for building and maintaining COBOL often shifts to outsourcers that employ developers in, for example, in India where the cost of hiring and training developers is lower. At the same time, a cottage industry made up of IT services providers that will migrate COBOL applications to cloud computing platforms using Docker containers has also emerged.
Many IT leaders are also concerned they will not be able to find enough IT professionals that have the IT operations skills required to manage mainframes. IBM along with other providers of IT management platforms such as BMC Software have been making a concerted effort to make mainframes more accessible to the average IT administrator by providing them with a set of graphical tools that function similarly to the tools many administrators employ today to manage distributed systems.
IBM going forward will also be investing in artificial intelligence to automate operations that previously required a lot of manual effort, says Barry Baker, vice president of product management for IBM Z and LinuxONE at IBM. “When it comes to AI there is much to be done,” he says.
Also read: PagerDuty Report: Stress on IT Teams on the Rise
IBM’s Mainframe Roadmap
Longer term, however, there are concerns about the mainframe processor roadmap. IBM in 2015 partnered with GlobalFoundries to build and design its next-generation processors. GlobalFoundries subsequently spiked both 10 and 7-nanometer processor projects that IBM was counting on to drive the wave of mainframe upgrades. IBM is now suing GlobalFoundries for roughly $2.5 billion in damages.
IBM also regularly reminds its customers that a mainframe is still the most secure platform available. Most recently, IBM extended support for encryption on the mainframe to new types of data sets, including sequential basic format and large format SMS-managed data sets. At a time when cybersecurity concerns are paramount, the ability to process data while it’s encrypted now resonates with organizations more than ever.
None of this means there won’t be organizations that decide to abandon mainframes for one reason or another. However, for the foreseeable future the number of organizations that will continue to rely on mainframe applications will result in the most venerable platform in all of IT likely still being employed well into the middle of the decade and beyond.
Read next: The Impact of Container Architecture on IT Development
