Performance
This topic describes frequently asked questions about performance and their answers.
- Is there a performance improvement when one converts from OS/VS COBOL to Enterprise COBOL?
- How do I know the performance gains after recompiling with Enterprise COBOL V6?
- What is the best way to measure the performance of a COBOL program before and after the migration?
Is there a performance improvement when one converts from OS/VS COBOL to Enterprise COBOL?
Yes. Enterprise COBOL V5 and V6 can give you a significant performance improvement when compared to all older COBOL compilers. It is especially true for programs with a lot of arithmetic.
For details about the key performance benefits and tuning considerations when using Enterprise COBOL, see the Enterprise COBOL for z/OS Performance Tuning Guide.
How do I know the performance gains after recompiling with Enterprise COBOL V6?
The only way to know the performance gains is to measure the performance before and after the migration. So if you want to find out what your performance improvement is, it is recommended that you back up your current Enterprise COBOL V4 modules before migrating your build compiler. When the migration, required recompilation, or optimization by IBM® Automatic Binary Optimizer for z/OS® (ABO) is done, set up a test environment, prepare the real representative workload, and measure the performance between the Enterprise COBOL V4 modules that you backed up and your new COBOL V6 or ABO-optimized modules.
What is the best way to measure the performance of a COBOL program before and after the migration?
- If it's a transactional application, you'll look at throughput for how many transactions it can process in a given amount of time, for example, transactions per second.
- If it's a batch application, measure the CPU or elapsed time of the batch job.
To measure the performance of a COBOL program before and after the migration, use a profiling tool like IBM Omegamon, IBM Application Performance Analyzer for z/OS, and other tools from IBM and other vendors. You could also use IBM RMF to measure job resource usage through the use of SMF records. For details on how to use RMF and other tools, see the white paper: COBOL Applications: Techniques to make them more Efficient. In performance tuning, you should find the hotspot and the particular modules that take the most amount of CPU time. It's not best practice to compare COBOL V4 performance before migration with COBOL V6 performance after migration because hardware, workloads, and code may all be changed during the migration. The comparison should be made on the same machine at the same time of day with the same data as input.