Mainframes are big, expensive and invoke memories of the enterprise as it was long ago. And yet, they are reliable, highly scalable and still form the heart of many organizations’ data environments.
The demise of the mainframe has been foretold too many times to count over the years, but as each new generation hits the channel, it somehow manages to maintain relevancy in the age of virtualization and the cloud. Without doubt, however, the nature of IT infrastructure is changing, and it is still not entirely clear that the mainframe will be able to change with it.
Of course, IBM has heard grim prophecies before, yet the company continues to put stock in its system z platform. The latest development comes in the form of a new software stack designed to help mainframe applications and workloads tap cloud and mobile resources. Enterprise COBOL for z/OS not only provides up to a 20 percent application performance improvement, but provides support for Java 7, UTF-8 coding and unbounded tables and groups for improved database flexibility. As well, it features new System Management Facilities (SMF) tracking tools that ease management requirements, and a new XML parser for data off-loading.
With nearly 15 percent of all business applications relying on COBOL in one form or another, it would seem that the mainframe is destined for enterprise environments despite the best efforts of distributed architecture proponents. However, software and hardware are no longer linked, and it is just possible that COBOL could reside quite nicely in non-mainframe environments – namely mobile. Micro Focus’s Ed Airey lays out a pretty strong argument that COBOL may even go native on mobile devices as users seek to replicate their office experience on the road. This would be a step up from current COBOL-based services in that it would maintain functionality when wireless access is down, and since it is such an elegant language, COBOL provides a top-notch modeling environment for high-performance, secure data delivery.
Still, the argument can be made that COBOL-based mobile applications would strengthen mainframe use in the data center, not weaken it. Perhaps, but what if an alternate technology were to come along that provided mainframe performance at lower cost and had the advantage of providing enhanced replication and backup services as well? Heirloom Computing says its ELPaaS system is already outperforming traditional mainframes on TPC-C and other benchmarks and can reduce costs some 90 percent. The platform uses 4-CPU/16 GB RAM resource blocks that scale dynamically and offer compatibility with IBM COBOL and CICS. The company says it can easily achieve 12,105 CICS transactions per minute – about the same as a 905 MIPS mainframe – running an IBM DB2 database aboard a VMware SQLFire fabric.
Indeed, as scalability and integrated systems become more prevalent, it is getting harder to tell what is and is not a mainframe anymore. Oracle’s Exadata platform, for example, consists of various processor, memory, storage and other modules, tied together with an InfiniBand fabric and aimed at massive workloads. On a functional level, it is a mainframe, but Oracle prefers the term “engineered system,” in that it uses mostly commodity hardware and can be tailored to either enterprise or cloud environments.
Ultimately, this becomes a matter of tomato vs. tomahto. Terms like mainframe and cloud are coined in the marketing departments of top vendors as a way to present their latest developments to the buying public. From a user standpoint, however, the focus should remain on solutions.
Regardless of whether it is a mainframe or a blade architecture, on the cloud or in the data center, at the end of the day, new technology needs to perform a vital function for the enterprise that deploys it. Finding the right solution at the lowest price point is the primary goal – and let the rest of the world worry about what to call it.