By: Megan Oster
Virtual tape has become a popular approach to storage because it features attractive capabilities of physical tape and disk in one solution. By provisioning inexpensive, high-capacity disk storage into logical entities, it emulates tape volumes. As a result, backup and recovery tools view virtual tape the same as they do physical tape, but virtual tape offers improved performance for fast recovery. This capability is important, as three major factors affect workload capacity performance: instruction path length, instruction complexity and memory hierarchy.
In order to complete a task, a job must execute a set of instructions. These instructions consist of various paths that run through the operating system, subsystems and application. The job path length is the total count of instructions executed across this system. The path length varies according to the complexity of the tasks, and the path length associated with the operating system can change based on factors such as competition with other tasks for shared resources and the number of logical processors of the image or LPAR. However, the application path length can remain the same in cases where the job performs the same task each time.
The type of instructions that a job completes, and the sequence in which it executes them, have a significant effect on instruction complexity. However, it is important to note that design can also impact complexity. The nearly endless list of design options includes instruction architecture, branch prediction and cycle time. Performance can vary between micro-processors in accordance with different designs, but it tends to remain similar across all models of that processor.
Typically, workload capacity performance is sensitive to how deeply into the memory hierarchy (the collection of caches, data buses and memory arrays that contain necessary instructions and data to complete a job) the processor must delve in order to retrieve instructions and data for execution. This factor affects performance because the further this information is located from the processor, the more time it takes to access the information; performance time varies in accordance with the average time it takes to retrieve instructions and data from within the memory hierarchy.
Given all of the intricacies involved in performance, leveraging a solution that can handle performance sensitive workloads well is critical. Fortunately, virtual tape is a suitable match for this challenge.