The Hitachi Unified Storage VM (HUS VM) can scale to 1,152 internal 2.5in/3.5in serial-attached SCSI (SAS) or multi-level cell (MLC) flash drives. It is built on a controller which can act solely as a storage virtualisation device if required and can manage a total of 64PB of internal and external capacity, including data on external arrays from other suppliers.
The hardware supports fibre channel and iSCSI block access protocols, as well as, optionally, common internet file system (CIFS) and network file system (NFS) file level access. There is no mainframe protocol support, which is present only in HDS’s high-end universal storage platform (USP) and virtual storage platform (VSP) enterprise SANs.
The HUS VM can also support object storage, which comes with the file access protocol option. But, object data can only reside in HDS object data formats and not in those of other suppliers. Currently, object storage systems are not standardised across suppliers.
The HUS VM is positioned between HDS’s HUS 100 mid-range storage array and its USP and VSP enterprise storage hardware. It is designed to compete with EMC’s VNX and VMAX, NetApp’s FAS 6000 and IBM’s XIV and V7000 lines.
HDS calls the HUS VM an “entry-level enterprise” storage array. According to HDS EMEA chief technology officer Bob Plumridge, the product gains its entry-level qualification due to pricing and ease of use, which is via the Hitachi Command Suite management suite common to all HDS storage arrays.
The HUS VM is a single-product model, with firmware unique to it and not interchangeable with the HUS 100 and USP/VSP lines. It is offered in three configurations: as a diskless storage virtualisation device for managing existing external storage, including from other storage vendors; with 2.5in or 3.5in hard drives and solid state drives; with a file storage and object storage option to provide multiprotocol storage.
An all-flash array option will be available in November, according to Plumridge. He said work was taking place to make the controller better suited to flash drive performance characteristics. These differ from those of spinning disk hard drives in terms of response time, and therefore bus bandwidth required and use of cache, as well as the need for features such as wear levelling.
“Storage architectures today are designed for disk drives, with read/write heads that move," said Plumridge. "We have guys working on developing the controller back end to optimise it for instant access and to speed data transfer. A lot of optimisation is needed to get the most from solid state drives.”