IBM and EMC are, along with HP, three of the main, if not the main, formers of opinion about storage virtualisation. This is by virtue of their market size, storage products, technical prowess and customer exposure. The three are heavily present in the SAN market and in enterprises and have talked to and listened to what their customers want from a storage provisioning and management standpoint. It is this set of needs that underpins virtualisation products which, certainly in the enterprise datacentres of the world, cannot be technology for technology's sake.
IBM was early into what we might call second-generation storage virtualisation with its SAN Volume Controller (SVC) and early into the idea of fabric intelligence with the arrangement to have it hosted on the CMS module in Cisco's MDS 9000 range of fabric directors. We can find out more about SVC here.
Although it runs in the fabric SVC is an in-band appliance, and therefore a potential traffic choke point due to scalability limits. We asked IBM if the storage virtualization market has moved away or is moving away from in-band appliances.
The company's spokesperson said: "No, IBM is seeing great adoption of SAN Volume Controller with more than 1,300 customers already deploying the technology. As we have demonstrated over the past two years, we continue to scale SVC in all the dimensions of storage; interoperability, scalability and performance. We plan to expand these dimensions again before the end of this year and will demonstrate unparalleled performance and scalability. This will be IBM's eighth release of SVC while EMC is still not present in the storage virtualization market."
In the Cisco director area a Fabric Application Interface Specification (FAIS) s being worked on to provide APIs for a storage array virtualisation and management product to interact with an MDS director. We asked if SVC supports it: "When FAIS becomes a standard, IBM will support it. Today, both Cisco and Brocade have submitted proposals for this standard but neither have been approved."
That's unequivocal. We also asked if IBM is considering using technology from Incipient to help achieve FAIS compliance. The reply was: "We do not discuss future roadmaps or plans."
IBM view of EMC's InVista
IBM is concerned to position SVC as market-leading and a more mature virtualisation technology than EMC's Invista. Another IBM spokesperson provided the following view:-
"It's been five years since the industry was gripped by the concept of virtualising storage environments to enable heterogeneous data sharing, improve utilization and drive down storage management costs. Since its inception, the race to storage virtualisation has taken off with thousands of customers world-wide readily embracing its streamlined management and cost saving advantages. But, as we count down the last days of third quarter 2005, something is amiss in the storage industry.
EMC originally promised it would deliver its storage virtualisation product, Invista, to market in August of this year. Today, after numerous reports and confirmations by the company, the product is still not generally available. Instead, only a select group of EMC customers will continue to trial the product, and volume shipments will begin in Q4. In short, it's taken five years for EMC to bring a virtualisation product to market. And it's still not here.
Looking back over this period, it's easy to understand why EMC has continued to slip on its promises.
- In early 2002, EMC told customers, reporters, and the industry that it already had virtualisation. Essentially, EMC was trying to convince the market that traditional disk array technology was equal to virtualisation.
- In January of 2003, EMC talked about virtualisation as "ill-defined'" and didn't like using the term.
- Four months later, in May 2003, EMC changed its position again and stated in an analyst report that it was going to address virtualisation and do virtualisation better.
- Fast forward to April 2004 and EMC tips its hand about plans for Storage Router at its annual users conference.
- One month later EMC states to stockholders at its 2004 shareholders meeting that, virtualisation "is probably the newest and hottest technology out there" and that they will have a storage virtualisation product - the storage router - in the second half of 2005.
- In March of 2005, IBM announces to the world that it now has more than 1,000 customers specifically using IBM storage virtualisation solutions across heterogeneous infrastructures - be it IBM, HP, Hitachi or EMC.
- On May 16, 2005, EMC introduces Invista at its Technology Summit in New Orleans. In the press release it states, "EMC will complete Invista beta testing this quarter, with general availability in the third quarter of 2005 for EMC Connectrix branded switches from Brocade and Cisco."
- Pause... and in August 2005 and EMC's vice president of infrastructure software, Todd Oseth, leaves the company to become COO at McData Corp. Oseth was responsible for the development of Invista.
For customers ready to embrace the virtualisation evolution, this timeline may warrant hesitation to engage with EMC. Not only has the shipping date been pushed back, but the company has been tightlipped about what functionality the product or switch will include.
So for more than two years, EMC has waffled on the need for storage virtualisation. During that same time, IBM delivered on products for storage, servers and software, and is now delivering its eight version of its SAN Volume Controller (SVC) virtualisation technology. Today, while EMC continued to evaluate and re-evaluate its single product virtualization strategy over the past few years, IBM has amassed more than 1,300 customers. And some of these customers are using SVC to virtualise their EMC arrays."
This is, obviously, a partisan view from IBM. No doubt a more technical feature-by-feature comparison between SVC and InVista will be possible once InVista's feature set is better known.
On the other hand
EMC, like IBM, is a successful and important storage hardware and software supplier. It has an impressive track record of delivering technologically innovative products to its customers. Its financial results are also impressive. It doesn't deliver dud products. Also it has apparently conceived, designed, written, beta-tested and delivered InVista in something like two and a half years. This could be viewed as excellent performance for a major, enterprise-class product, even one that could be presented as a hurried catch-up product as IBM is doing. We'll look at aspects of its virtualisation technology in a future article.