The mainframe has always been able to support complex storage environments and every type of application from mission-critical to archival applications. In contrast, non-mainframe systems (UNIX, Linux, and Windows) were designed for computation and not to perform intensive data and storage management tasks. No one knew then that these computing systems would one day be asked to do the work of a mainframe. As a result, average disk allocation levels for UNIX and Windows systems average only 30 to 45 percent of total drive capacity while mainframe disk storage averages about 80 percent allocation. Somehow, this low level of storage resource utilization continues to be tolerated by non-mainframe businesses and contributes heavily to storage inefficiency while adding cost.
In 1976, IBM’s Hierarchical Storage Manager (HSM) software quickly gained appeal for its ease-of-use and automated, policy-based capabilities that optimized storage resources. Today, most mainframes run HSM on a 24x7 basis, continually migrating less active data to more cost-effective storage devices, typically tape, freeing expensive disk space for more important applications. As a result, mainframes have much larger tape environments than open systems. Though a few HSM-like software products exist for non-mainframe systems, deployment isn’t widespread, thus, allowing low-activity data to accumulate on disk and increasing hardware and energy costs.
In April 1988, IBM announced DFSMS for its mainframe computers. Commonly called “SMS,” this architecture consisted of a set of related software products that marked the most comprehensive set of policy-driven storage and data management capabilities for disk storage introduced to date. More than 20 years later, non-mainframe systems still don’t have storage management solutions comparable to HSM or SMS.
Mainframes use tape for a wide variety of applications, including back up/recovery, production files, and archives. Non-mainframe systems have yet to advance their view of tape much beyond backup/recovery and still use disk for archiving large volumes of low-activity and archival data. In 1998, IBM and StorageTek pioneered integrated Virtual Tape Libraries (VTLs) on the mainframe. Integrated VTLs include a robotic tape library integrated with a disk array, combining tier 2 and tier 3 solutions into a single architecture. Integrated VTLs reduce the number of physical tape cartridges by placing multiple, virtual tape volumes on a single physical cartridge. Tape cartridge utilization commonly reaches 80 percent or more compared to much lower average utilization levels without integrated VTLs. By 2008, 10 years after integrated VTLs were introduced to the mainframe market, integrated VTLs began to enter the non-mainframe market as tape systems struggled to advance beyond just backup/recovery and into true tier 3 applications such as fixed content, compliance, and archive.
Mainframes Achieve the Highest Storage Utilization Levels
The “triple threat” combination of HSM, SMS, and integrated VTLs enable mainframe storage environments to consistently achieve the highest utilization levels for disk and tape and to deliver on the promise of tiered storage. Several non-mainframe disk suppliers are now touting the concept of “thin provisioning” as a breakthrough to improve low levels of disk utilization. However, this concept isn’t new and first appeared with the JCL SPACE parameter (as the Release subparameter [RLSE]) in 1965 on the OS/360 operating system. Thin provisioning will help improve the low-utilization levels that have always plagued non-mainframe systems.
Storage capacity continues to grow faster than the deployment of management tools, and the supply of skilled storage people lags behind the demand. This is especially true for non-mainframe systems. Extensive staff downsizing has impacted many organizations, leaving the remaining personnel focused on keeping the IT function running. By year-end 2010, it’s projected that a non-mainframe storage administrator should be able to manage an average of 30TB of disk storage. In comparison, the typical mainframe storage administrator using powerful tools effectively manages more than 100TB of disk storage. Remember, these are averages and can vary considerably based on skillsets and management tools available.
It’s an ongoing goal of non-mainframe systems to deliver a high level of storage management capabilities comparable to the mainframe. Approaching this goal appears unlikely in the foreseeable future and these systems grow much faster than their storage management capabilities. Most non-mainframe storage management solutions come from a variety of suppliers, making interoperability a challenge, and require a customer to coordinate with many vendors. The era of combining weak operating system storage management services with increasingly large storage subsystems needs to close as inefficiencies are increasingly costly. But will it?