IBM's new System z13 mainframe has been the subject of recent posts appearing on this blog. "15 Reasons IBM's z13 Mainframe Might Just Succeed" laid out several prominent reasons the z13 would appeal to large organizations. "Is the Buzz Surrounding IBM's System z13 Mainframe Just Hype", which featured veteran system programmer, Eloy Cruz-Bizet took, an initial look under the hood of the z13 Mainframe. This post will take yet a closer look under the hood at language-specific enhancements available in the z13, and what those enhancements may mean.
The implementation of performance and capacity improvements on a hardware level, such as those introduced with a new model mainframe processor such as the IBM z13, must take into account concerns of maintaining compatibility with preexisting application code. In addition, a major endeavor must be undertaken by the hardware manufacturer to maintain transparency with existing program code to prevent disruption of the existing customer base of in-house written software.
Contrast this, to the efforts which may be required by a particular firm anxious to actually take advantage of and leverage performance and capacity improvements offered by the changes to, or acquisition of new hardware. Hardware manufacturers can only go so far in making improvements before the hardware becomes borderline incompatible with the base of existing in-house written software. Invariably, hardware evolution reaches a point where the new feature requires alteration of the microcode. The roll out of the z13 has introduced 660 additional op-code machine instructions .
At some point such hardware progress will require the source code to be recompiled as a method of introducing new features into the compiled code, by means of production compiled code with the machine language references and access to the new hardware features. Since knowledge of these microcode improvements did not exist at the time the current base of applications were last compiled, IBM is forced to implement software access to these hardware features within the scope of concurrently releasing improvements to the range of compiler software offerings, as well as macros, and calls for CICS, DB2, TPF, etc.
The most prominent and persistent languages found in application programs running on the mainframe are COBOL, PL/1, C/C++, and the High level assembler. Billions of lines of code have been written in these languages. This code forms the foundation of many companies’ core applications such as payroll, human resources, general ledger, insurance, accounts management. Embedded within such code are also often the rules, both regulatory and business proprietary which, drive much of these applications, and indeed, drive the business itself. Once the dust of Y2K remediation settled, the existence, depth and often complexity of these rules was brought to light in a most prominent fashion. The dawn of fear in the mainframe world gained a foothold in business during this period, as did the realization of the impact of application programmers moving to other companies or retiring and in doing either, taking intellectual property with them in their heads.
Consequently, not all companies will jump on the legacy modernization bandwagon, due to fear of not being able to replicate the same result once the modernization has been completed. Only 48% of the 1,194 respondents to BMC's 2014 Annual Mainframe Research Results published in October 2014 had mainframe modernization as a top priority for 2015. Inherent in such modernization is an element of risk which most find uncomfortable. There is also the cost of finding and hiring qualified talent to understand and then translate the business and regulatory rules mentioned above. Although much progress has been made from the days of spaghetti code, there is no doubt in this writer's mind that it still exists.
However, one of the z13's big selling points to address this concern is the enterprise COBOL for z/OS V5.2 compiler which includes application modernization as a feature. Astute leaders of mainframe applications will nonetheless establish baselines and perform extensive testing to assure the same or better results as those produced by the legacy application. With remediation, cost remains and additionally, a new element of risk related to compilation is introduced into the process. This is especially true where the COBOL program makes calls to or links with external sub-programs. The remediated COBOL program may compile successfully only to have the job abend due to unexpected incompatibilities in the static re-linked, dynamically called system routines or called sub-program's load module. Although, IBM claims that any necessary changes may be introduced piecemeal by the use of selective recompiles, changes and reassembly in lower-level routines, I/O calls, database calls, etc. may also be required to establish compatibility with the primary identified COBOL, PL/I or C/C++ program. This can create a chain-effect of needing sub-compile and reassemblies. Furthermore, for those firms doing assembler language programming, there is a fear that IBM’s choice of 660+ new Assembler language instructions mnemonic op codes for new z13 hardware features, will conflict with similarly named user-named in-house application macros. At that point the ability of programming department management to estimate the budget of an application software performance improvement plan, let alone estimate the magnitude of liability risk with it, become daunting.
Summary
In summary, there are many features and enhancements available in the new IBM System z13 mainframe. Inherent in the decision to include such a purchase must also be a realization that to fully exploit the performance metrics frequently touted by marketing, extensive planning must be incorporated into the overall upgrade plan, along with a buffer to accommodate the unknown which will inevitably arise. This is not a SAAS solution where commercial software will work 'out of the box." As such, it is not a quick fix to what ails you, but a calculated journey to entering the competitive multi-billion dollar mainframe mobile app market. It is a journey which is not for the faint of heart!
Brenda J. Christie
Sources:
http://www.ibm.com/support/docview.wss?uid=swg21693594
http://www.ibm.com/support/docview.wss?uid=isg1PM79901
http://www.ibm.com/support/docview.wss?uid=swg21694301
No comments:
Post a Comment