HDS blows performance socks off storage competition, intros SOA for storage

Written By:
Content Copyright © 2007 Bloor. All Rights Reserved.

In a raft of announcements, Hitachi Data Systems (HDS) has introduced a new intelligent storage controller that moves storage capacity and performance to new levels, extended thin provisioning and introduced a service oriented storage solutions (I’ll call it SOSS) architecture supporting service-oriented architecture (SOA) and charging according to storage usage.

The capacity and performance figures are so far in advance of anything currently available that they take credibility to the edge. So, for instance, the massively parallel Universal Storage Platform (USP) V controller offers 3.5 million I/Os per second (IOPS), 320 concurrent internal memory operations, 106 GB/sec aggregate internal bandwidth, 247PB total storage capacity (332TB internal), and connectivity to 224 FC, 112 ESCON or 112 FICON ports.

Few would argue with Hu Yoshida, Hitachi VP and CTO, who said today that the new universal controller “will blow the socks off the industry”. It will certainly be interesting to see how major competitors like EMC and IBM will respond. Hoshida’s slightly O.T.T. claim was that the USP V “has no limits as its only limitation”. (I do see what he means.)

More practical is that this available performance and capacity boost stands to ‘re-energise’ existing storage hardware. Thanks to heterogeneous connection, HDS also claims the world’s first implementation of a large-scale, enterprise-class virtualisation layer, and it uses this for thin provisioning with logical partitioning (using Hitachi Dynamic Provisioning software) to allocate only the amount of requested storage actually needed. This will also simplify data replication for a mix of storage hardware—which Hoshida described as producing a “liquid pool” of storage. HDS’s existing HiCommand storage management software manages the mix of vendor hardware and software regimes.

With SOSS, Hitachi is introducing SOA concepts to storage, providing a set of automated functions delivered as services that can be invoked when needed. This has two key potential benefits, one flowing from the other. The first is the ability to implement realistic storage charge-back schemes instead of some vague cost apportionment between departments or partner businesses; so lines of business (for instance enterprise sub-divisions) can focus their minds on provisioning storage according their needs.

The second follows the path of SOA by allowing large legacy storage solutions to be deconstructed into loosely-coupled sets of services to be invoked by vertical applications when needed. Examples of these include: business continuity services, content management services involving search and indexing (using Hitachi Content Archive Platform), non-disruptive data migration and storage security services. These can be presented as a menu to business clients (including internal ‘customers’), with charging according to what is selected and used. SOSS can also be used to ensure QoS and service level objectives and agreements can be met.

This is clearly very high-end stuff. HDS is promising cut-down controllers for the medium and low-end markets, but would not provide time-scales for these. Meanwhile, thin provisioning and the new controller design plays well in the ‘green storage’ marketplace with reduced power and cooling requirements—presumably until you take advantage of all that extra performance and capacity to do things you never dreamt of doing before.

Large enterprises, who all have a varied mix of storage hardware, should certainly be taking a close look at what HDS is offering today.