If you’ve ever had an application terminate unexpectedly or a mainframe operating system suddenly stop, it can be a frustrating experience, especially in the wee hours of the morning. These abends often occur when there are errors or conditions present that the application or operating system cannot manage.
During the Aug. 7 “Panel Discussion — A Review of Ways to Prevent Out-of-Space Abends” at SHARE Kansas City (Aug. 4-8), Russell Witt, senior consulting software engineer at Broadcom, will moderate a panel of experts, including Tom Conley, senior consultant for BMC Software, and Jerry Spencer, mainframe systems engineer at DTS Software, on products that can help mainframers prevent out-of-space abends and other allocation issues, as well as answer any questions about those products.
According to Spencer, “One issue is the difficulty of specifying 'correct' values for primary and secondary space. Given that both the amount of space required and the amount of space available may vary widely, users tend to request substantially more space than is needed, just to be on the safe side.” Architectural limits in z/OS, such as “as those on the number of extents (e.g., only 16 for 'garden-variety' sequential datasets), and dataset and volume size limits” can cause issues. “The difficulty of changing job control language (JCL) and application parameters also contributes to inefficiency and errors,” says Spencer. “The ideal solution would be a means to make application allocations more centrally and dynamically controllable.”
Out-of-space abends and JCL errors can occur if the requests for primary space cannot be satisfied. Spencer points out, “Similarly, and even more commonly, secondary space and volume-addition failures lead to application failures and time and resource-consuming reruns.” He adds, “When too little space is requested, excessive extents can again lead to poor performance or application failures.”
Conley indicates that storage tools, such as BMC AMI Storage, can provide a full set of controls to address allocation issues at several points of control during the allocation process (e.g., extending onto another volume and reducing primary/secondary space). During the session, Conley says he will outline the points of control and show examples for how to address issues at those control points.
According to Spencer, IBM’s Data Facility Storage Management Subsystem (DFSMS) offers several tools to address allocation issues from data class definitions, naming standard enforcement, and automatic class selection routines. “An ideal solution would be one that provides simple IF/THEN/ELSE logic to control allocation,” he explains. “By interceding just before the failure occurs, applications are no longer subject to failure. Vendor products, such as DTS Software's Space Recovery System (SRS) and Allocation Control Center Monarch (ACC), are extraordinarily mature and robust, providing much more detailed error prevention and control over allocation than is available in just z/OS and DFSMS. It’s been said that ‘hope is not a strategy,’ and now both operations and the production control team don't have to cross their fingers and hope that everything will run successfully.”
Another example of an abend that can cause delay is the X37 abend. Conley explains, “An X37 abend will stop a job in its tracks, requiring manual intervention to modify JCL, initialize more volumes for a pool, process DFSMS Automatic Class Selection (ACS) routine changes, etc.”
To modify JCL in the event of an X37 abend, he explains “it is often an iterative process until you find the exact correct SPACE parameters that successfully allocate the dataset.” Conley adds, “In extreme cases, an X37 abend outage could impact service level agreements. With BMC AMI Storage, mainframes can avoid prolonged outages by automatically recovering from X37 abends. This keeps the job running and prevents disruptions in operations.”
For non-SMS datasets, mainframers still need a product like BMC AMI Storage to provide recovery for X37 abends, the ability to direct non-SMS allocations to a specific pool, provide more control over SMS allocations, and more, according to Conley. “Recovering from StopX37 abends allows our clients to maximize their storage resources by ensuring that any inhibitors (e.g., poor JCL space parameters or fragmented pools) are handled automatically without requiring manual intervention,” Conley says. “This reduces the workload on both storage administrators and mainframe operations by eliminating abends, which allows job schedules to run smoothly without outages.”
Attendees at SHARE Kansas City will learn how space error prevention and allocation control can become a form of automation for z/OS. Spencer adds, “As the one who was called at 3:00 a.m. because of a B37 abend, I can testify that it is a wonderful thing when the system handles the issue, and your slumber is not disturbed.”
Register today for SHARE Kansas City.
Serena Agusto-Cox has more than 20 years of editorial experience and six years of experience writing about mainframe and information technology. She interviews and crafts forward-looking and engaging technical updates related to the mainframe ecosystem, highlights the experiences of thought-leaders in the community, and shares important updates to technical education and training.