Data Standards Enforcement for Mainframe Reliability and Cost Savings
The world’s most vital enterprise IT systems still depend primarily on mainframes.
By Steve Pryor ,
DTS Software, LLC
Among the mainframe’s advantages that make it so valuable are its centralized control of data sets, applications, and devices and the facilities for managing them. Large-scale data set allocation software such as DFSMS, disk management systems such as DFSMShsm and FDRABR, and tape management systems such as DFSMSrmm and CA-1 provide a level of efficient, policy-based management that users in non-mainframe environments can only dream of. A community of ISVs supplement the mainframe’s usefulness by providing both systems and application software for z/OS. Many of these ISV products are aimed directly at saving costly computing resources such as I/Os, CPU consumption, elapsed time, disk space and so on.
One important means of saving resources is through the establishment and enforcement of data center-wide standards. Standards make it possible to classify data and bring order out of complexity. With proper standards in place, data arrives in the proper form, when and where it’s needed, and processes operate at the right time with the right I/O and resource use is minimized.
Arguably, the most important of mainframe data center standards is a set of logical and consistently applied data set naming standards. Much systems software relies on the format of the dataset name to determine what to do with the dataset. The storage management subsystem, DFSMS for example, uses the Automatic Class Selection (ACS) routines to provide policy rules that assign SMS classes (the data, storage, and management classes) that determine how datasets will be handled during their life cycle. While the ACS routines can examine many different dataset characteristics, in most installations ACS routines are coded to rely on data set name patterns when assigning classes.
Similarly, disk management software systems, such as DFSMShsm and FDRABR also tend to rely heavily on the use of dataset names (as well as disk volume names or storage group names) to determine when datasets and volumes should be backed up, how long backups should be kept, and when to move datasets to archive storage such as tape. Naming conventions, particularly for disk datasets, are thus critical to efficient data center operation.
Standards to Address Legacy Mainframe Systems
Besides data set and volume names, other important mainframe data center standards include those related to JCL and control statements such as those for IDCAMS or other utilities. Because JCL and control statements describe how datasets are to be defined, opened, and accessed, consistency in these standards is just as important as in dataset naming, albeit more difficult to achieve. Unfortunately, while DFSMS does an admirable job of enforcing basic policy standards based on dataset name, there is no centrally administered enforcement mechanism in z/OS for JCL or control statement standards.
For example, many installations are saddled with aged, obsolete JCL written at a time when certain resources, such as virtual storage, were very limited, and they rely on buffering defaults (e.g., BUFNO, BUFNI, BUFND, etc.) that have not changed since the 1960s. The result today is poor performance and excessive I/O and CPU consumption. Similarly, because JCL and control statements are static and seldom changed, installations often fail to take advantage of new technologies such as VSAM System Managed Buffering or extended-format (greater than 255-generation) GDGs.
There are many other instances where standardized management of JCL and control statements can provide consistency and reduce error—in allowing, preventing or overriding EXPDT and RETPD values, for example, ensuring that the VSAM Record Level Sharing (RLS) values assigned to a dataset (such as logfile name) at DEFINE time are appropriate, based upon whether the dataset is to be used in production or test. This sort of functionality isn’t available in the base z/OS operating system, but fortunately ISVs provide software which can receive control and apply such standards whenever are datasets created, referenced, opened or used. This robust software has been around for years and can eliminate the need for manual or mass JCL changes as well as provide SMF data, logging and reports.
The Mainframe’s Reliance on Standards
Standards, whether for data set or volume names, JCL, control statements, program or device names or other functions, are at the heart of the reliability and usability of mainframe systems. Standards allow for the most efficient use of computing resources, as well as the efficient use of valuable programmer/analyst time and effort. Engineering resources–people–are the most expensive and limited resource in any data center. The common, centralized control provided by standardized policies in z/OS, including ISV products, is key to making the people supporting the mainframe the most valuable and efficient resources in the computing world.
Steve Pryor is DTS Software's Chief Technology Officer. Steve has a more than 30-year background in storage management, design, and support, and speaks frequently at industry events.
See more by Steve Pryor
About DTS Software, LLC
DTS Software is recognized worldwide as the leader in enterprise storage management software. Learn more about our solutions →