Contents |
Anouncement Summary
Hitachi does not usually put on a big event for their announcements. With the Hitachi announcement of the Virtual Storage Platform (VSP) in Santa Clara, Hitachi celebrated both 100 years in business and spelt out a clear vision of the data center storage infrastructure requirements over the next few years. It enunciated a storage scale-up strategy, a scale-out strategy, a scale-deep strategy and a storage management strategy. It scored big on scale-up and scale-deep, and the jury is still out on scale-out and storage management.
Scale-up Storage
Hitachi is at heart a tier 1 storage supplier, where tier 1 is defined as the very highest in performance, the greatest ability to make a rogue application sing, and the very highest in business continuity in all its aspects. The heart of the VSP storage system is a very high performance ESW switch built from PCIe-based interconnect components. Hitachi as taken SAS 2.5 inch drives and can supply the I/O requirement for more than 2,000 drives. Although benchmarks have still to be completed, this compares very favorably on paper with the more distributed architecture of VMAX based on a RapidIO fabric, and the more limited scale of the IBM 8000 series. Other 1.5 vendors such as 3PAR, IBM XIV and NetApp give good enough performance without the ability to drive real low latency for specific applications, and they do not have the track record for availability that makes EMC, IBM, and Hitachi stand apart. Table 1 summarizes the positioning for Tier-1 Scale-up storage.
Scale-out Storage
Hitachi has approached scale-out by making the different components of the system individually scalable. More front-end, back-end, cache, or virtual storage processors can be added independently to tune the system to any workload. In this area EMC’s use of standard components and distributing up to eight sets of processors connected by RapidIO scores as best of breed, with the theoretical ability to stretch this to even more engines. Scale-out however only works when you have workloads that are relatively well mannered and can run within their own zone. IBM does not have a scale-out strategy with the 8000 series. Table 2 below summarizes the relative positioning for Tier-1 Scale-out storage.
Scale-deep Storage
Hitachi has a very strong storage virtualization story with the VSP, building on the virtualization capabilities of the USP V. IBM also has a strong and very functional virtualization story with the SVC, which is focused more on the mid-range market. EMC has a new strategy with VPLEX, which is a very innovative high-performance distributed architecture but lacks the base functionality of either Hitachi or IBM. Only Hitachi has a single box and architecture that provides the virtualization and the array functionality combined in a single platform. Table 3 below summarizes the relative positioning for tier 1 scale-deep storage.
Storage Management
All the tier 1 vendors have developed strong storage management stories. Hitachi announced High Command V7, with improved ease of use and an overall clear vision to provide an integrated platform for the lifecycle management of all data types. Gary Sloat from BB&T gave a warm endorsement of Hitachi’s VSP, but put his finger on the problem that all storage management vendors have in this space:
- Application developers never architect lifecycle management into applications because of the pressure to address today’s problems and implement applications early.
- Retrofitting lifecycle management with automated classification, metadata analysis, or magic has enjoyed limited success to date.
All vendors will struggle with an overall solution, because this problem cannot be solved by a storage-based approach alone. Developers and ISVs would need to embrace an open standard framework before there can be much movement, and that is unlikely to happen soon.
Summary and Conclusions
There are three very different strategic approaches to providing a unified storage strategy for Infrastructure 2.0. IBM offers many very different storage products and provides a virtualization umbrella with SVC. EMC is attempting to integrate a Tier-1 scale-out VMAX architecture and a separate VPLEX virtualization infrastructure. Hitachi is integrating the array and virtualization engines. As a result of the announcement, Hitachi has significantly improved its position in the Tier-1 marketplace, and depending on pricing, offers a good Tier 1.5 product as well.
Action Item: Hitachi has laid out a well thought-out strategy for Storage Infrastructure 2.0 with excellent products for scale-up and scale-deep, and a much improved storage management product and framework. Hitachi’s customers should be pleased with the VSP announcement, and Tier-1 sites without Hitachi storage should re-look at Hitachi’s Tier-1 portfolio.
Footnotes: