EMC, an industry leader in data storage, recently introduced technologies that give mainframe storage users much greater ability to actively manage the performance of stored data. As explained by EMC’s Flavio Fomin, the new technologies are Virtual Provisioning (VP) and Fully Automated Storage Tiering (FAST) VP. In an interview with Enterprise Tech Journal, he explains what these technologies do, how they work, and what benefits they provide.
Fomin is a member of EMC’s Symmetrix Engineering group. His current focus is in Innovation and Systems Engineering and he’s one of the members of EMC’s FAST Project. A 12-year EMC veteran, Fomin has handled different engineering and line operation positions, supporting the Enterprise Storage business. His contact with the IBM mainframe started in 1984, when he learned Assembler programming in Brazil, a skillset he has always found valuable. Fomin has several years of experience in capacity planning and performance, and over the past 12 years has introduced several performance modeling tools and products. He has a bachelor’s degree from Faculdades Associadas de Sao Paulo, a master’s degree from the University of Sao Paulo, and an MBA from the MIT Sloan School of Management.
Enterprise Tech Journal: As a leading storage vendor, EMC often has something new to introduce. What’s happening now?
Flavio Fomin: In May at EMC World, we introduced both virtual provisioning and fully automated storage tiering for Virtual Pools (FAST VP) for the mainframe storage environment.
ET: Can you explain virtual provisioning, or resource deployment?
Fomin: It’s a method of provisioning that enables a more effective use of storage capacity. Instead of provisioning storage on a single disk drive or RAID [Redundant Array of Independent Disks] group, virtual provisioning disperses data in small, 12-track chunks across all disk drives that are made available to the virtual provisioning pool. These data chunks are wide-striped across all disks in the VP pool, providing better workload balancing at no additional cost of management.
The more efficient use of the storage capacity is because, with virtual provisioning, the space is allocated only when it’s used. For example, if a data set uses only half the allocated capacity, the other half could be reclaimed and used by other data sets/applications.
Some sites prefer to pre-allocate the entire space required by the volume being provisioned. In these sites, the benefit of virtual provisioning is workload balancing. Other sites are interested in both “over provisioning” and better workload balancing.
ET: How is virtual provisioning different from traditional mainframe provisioning?
Fomin: Virtual provisioning doesn’t change how storage devices are presented to a host computer. If you’re starting from a reference point of traditional mainframe provisioning and moving to virtual provisioning, making the move is entirely transparent and virtual provisioning can be easily implemented. Virtual provisioning introduces changes to how the storage administrator manages the storage array. The idea is that the storage administrator will manage the virtual provisioning pools instead of managing the individual disk drives that are part of the pool.
Virtual provisioning is a first step toward better storage management, but when you combine virtual provisioning with FAST VP, you have active performance management of the data at the sub-volume and even sub-data set level.