COBOL has just turned 60 this year and it is still running the world’s most critical business applications. At IBM, COBOL and Z processor development teams work in close collaboration to jointly deliver innovations to improve the performance of COBOL workload on modern Z mainframes. This approach to leverage advancements in both hardware and compiler technology to maximize the return on investment is unique in the industry.
What’s new and exciting in Enterprise COBOL v6.3 (COBOL v6.3) and Automatic Binary Optimizer v2.1 (ABO v2.1)?
Both offerings fully exploit the new Vector Packed Decimal Enhancement Facility and significantly boosting the performance of business-critical COBOL applications running on z15. Compared to the same applications compiled with COBOL v4.2 (or below), compute intensive applications compiled and optimized by COBOL v6.3 reduces CPU usage by an average of 58%; and binaries optimized by ABO v2.1 reduces CPU usage by an average of 57% on z15. ABO v2.1 has now achieved close to on par performance with COBOL v6.3. Also, COBOL applications encountering decimal overflow conditions while using interlanguage calls (e.g. COBOL calling Java, PL/I or C/C++…), could get up to 20% and 22% performance improvement with COBOL v6.3 and ABO v2.1 respectively. The performance impact could be significant for applications that frequently encounter these conditions. *
COBOL v6.3 has a number of new features designed to help developers modernize core COBOL applications to support digital transformation. They include:
• Native UTF8 support
Developers can now build applications that process UTF8 directly. This will greatly improve the efficiency of processing national languages and reduce development effort.
• New COBOL language features to improve flexibility and reduce verbosity
Dynamic-length elementary items
FUNCTION keyword in REPOSITORY paragraph
• COBOL 64-bit language features (first delivery)
Allows developers to create batch applications to access one or more data items up to 2GB each without changing program logic or developing complex ”data-windowing” techniques to manage large tables
ABO v2.1 optimizes programs built with CMPR2 option in VS COBOL II. This increases the increasing opportunity for the optimizer to improve application performance. There is also improvement in scalability. Optimizing large modules with ABO v2.1 can get up to 57% reduction in CPU and 30% reduction in memory usages over ABO v1.3.* In addition, ABO v2.1 now uses Execution-based licensing terms (instead of z/OS based terms). This offering is now priced base on the utilization capacity of the LPARs in which ABO v2.1 actually runs on, to provide pricing flexibility. ABO v2.1 generates SMF-89 records to enable clients to manage utilization capacity.
ABO v2.1 is designed to complement COBOL v6.3. Many development organizations only compile code that require source changes. By using ABO together with the COBOL compiler, clients can compile programs that require source changes with COBOL v6.3 and optimize modules that are not frequently updated with ABO v2.1. This way, CPU utilization and the operating costs of business-critical COBOL applications can be reduced to maximize ROI without a major impact to current development process.
COBOL v4.2 is scheduled to go out of service on Sept. 30, 2021. Many companies have either started or in the middle of migrating to COBOL v6. The effort required for migration depends on the number of issues found in existing source, as all of them would need to be fixed. ABO v2.1 can help significantly reduce migration effort. Clients only need to migrate code under active development or maintenance to COBOL v6.3, and then make it the default production compiler. They can use ABO v2.1 to optimize performance critical modules that were built a long time ago. Because ABO directly optimizes working binaries, there is no need to detect, analyze, test and fix issues in existing source. Binaries optimized by ABO v2.1 now have similar performance improvements as code optimized by COBOL v6.3 on z15. As a result, clients can significantly reduce migration effort while improving application performance. Best of all, they can do it without any major impact to their development process.
market place pages to learn more about these cutting edge offerings.
* Performance results shown was obtained in a controlled, isolated environment using IBM internal test suite. Performance of other workload may vary