The phrase “managing storage” used to mean that a team simply managed the storage infrastructure. At that time, DASD storage was managed at a capacity level and pre-RAID mainframe disks spun at the same speed and were the same size in raw disk capacity for each model type. In addition, cooling and power considerations were only bit players when factoring in cost and productivity. In this era, storage was all about capacity and whether there was enough to get through another night of batch processing. There were plenty of skilled storage management personnel to manage the capacity without sophisticated tools.
Storage Resource Management (SRM) software became an integral part of managing storage resources as RAID technology emerged. These products assisted in managing the newer technologies and growth as storage disk hardware costs began to rapidly fall. Even with these tools, the emphasis was on storage capacity. The real problem for storage management data continued to exist and with it the need for cost-effective management.
The mainframe’s ability to provide cost-effective, stable computing makes it an excellent choice for critical applications in today’s economic conditions. With an estimated 60 percent of all storage-related spending tied to the total cost of managing the world’s installed base of external storage, storage managers can lead the way in reducing the overall costs. The challenge will be determining which methods are to be used to maximize existing resources.
Changing Our Ways
Analyst estimates show that unused or unavailable RAID overhead consumes 20 to 50 percent of the installed disk storage capacity. Statistics show that the amount of mainframe data is doubling annually, so there’s a significant amount of wasted disk resources. Even with RAID subsystems, the cost to power and cool the spinning drives can be exorbitant. Investing in newer hardware technology won’t remove the waste.
The leading cause of wasted space is the lack of understanding of JCL space allocation, coupled with the assumption that DFSMS does correct allocation based on what is needed. Application programmers must determine the correct amount of space to allocate, so quite often there are incorrect requests. The net result is large amounts of wasted idle space that no one else can use.
There are solutions that allow technicians of various skill levels to identify areas of data inefficiencies and automate optimization. Higher utilization of existing resources delays capital expenditures while ensuring the validity and availability of the data.
Achieving the End Result
IT requires cost-effective solutions and the right mindset to manage at the data level. Some best practices that will provide data management cost savings include:
• Reducing idle space: Using correct allocation metrics or an SRM solution to adjust the allocation based on what is needed—not what is requested—can provide tremendous savings.
• Understanding the access patterns of the data in place: Periodic reviews of the DFSMS management classes and the type of data that’s being assigned to the management classes should be conducted and reviewed with application teams. If the data is infrequently used, move it to alternate media or delete it.
• Using virtual tape to house infrequently used disk data: Define policies that move infrequently used disk data to virtual tape volumes, freeing up precious disk space.
• Establishing accurate forecasting of data allocation trends to delay or prevent unnecessary DASD purchases: Preventing unused drives from spinning provides cost savings for power, cooling, and unused resources. Use of application groupings to trend data use over time at the application level provides an accurate determination of how much additional storage is needed and when.
• Using tiered storage to reduce the amount of power used on faster spinning drives: Tiered storage plays a large part in how and where the data is placed based on performance metrics. Faster spinning drives are usually smaller, use more power, and produce more heat requiring added resources to cool and power the equipment. Slower spinning drives optimize the physical capacity by allocating non-critical performing data onto larger physical disks.
Managing at the data level reduces costs associated with power, cooling, administration, and hardware and makes the existing resources pay for themselves. To achieve the savings and prevent unnecessary hardware purchases, use appropriate storage resources for the type of data used. This removes wasted idle space and maximizes the existing hardware in place across the entire mainframe infrastructure.