JCL: What is virtual TAPE and how it works

Virtual Tape is a software product which intercepts tape mount requests, and supplies a 'virtual tape' instead. The virtual tape is actually a dasd buffer - the software processes all tape I/O requests through the dasd buffer.

In a write secenario, the dasd buffer (hopefully large enough to hold the entire dataset written to tape) is later staged to a real tape device (cartridge, magstar, etc). An index mechanism keeps track of where this 'virtual dataset' is on the real tape device. In this way, the application program gets maximum performance, and the tape devices are fully utilized (each is 100% full, instead of the typical 2%).

In a read scenario, the real tape device holding the required dataset is mounted, the dataset is copied to a dasd buffer, and the all the read requests are satisfied from the dasd buffer. This typically imposes a slight delay while the tape device is mounted and then positioned to the requested dataset. Of course, if the dasd buffer has not yet been staged to a tape device, then there is no need to mount and position it, and the read requests are immediately satisfied.