Availability, operation expense and performance compose the triangular arena of requirements for running the larger than ever and ever-growing mainframe database systems nowadays. Availability should be continuous, operation expense minimized, performance satisfying. All requirements do contradict each other, so balancing the triangle and finding the compromise in-between is a non-trivial task. Acquiring a state-of-the art technology raises costs, performance tuning inquiries may affect availability, too much cutting back on the budget will endanger both availability and performance. And keeping a flock of experts busy with performance and availability issues is mostly not feasible in these times of desiccated staff resources. So ever too much the triangle gets out of balance, in favour of the lowest of all possible operation expenditures. That may be dangerous.
So why not let the machines care for themselves? Find a capable toolset that analyses I/O performance and develops tuning measures, based on its built-in expert system. Solutions must be presented in a way that staff accepts them and implements them, nothing else. Use an application that acts as an automated whistle-blower, highlighting availability issues far before they actually become a problem. So that staff can do their daily work, rather than acting fire fighters. Establish an analyzer application for the ubiquitous, resource devouring dynamic SQL. Out of millions of SQL statements, it will comfortably separate the wheat from the chaff and show where and how performance can be easily improved.