So far the smart-switch based product has failed to set the market on fire. By February this year – well over a year after the product was launched – EMC was claiming only something over 100 customers, and at that stage the company would not say how many customers had put the system into production use.

Now EMC says that Invista has been taken up by around 200 customers, and that much of the boost to the sales rate has resulted from the release last month of version 2.0 of the code, which runs on smart SAN switches sold by Cisco Systems and Brocade Communications.

EMC has not publicized the release of Invista 2.0, but it says that the upgrade has not gone un-noticed by its customers.

We’re seeing a big uptake largely based on the 2.0 release, said Doc D’Errico, EMC’s infrastructure software group vice president and general manager.

Although EMC said that the upgrade improves the performance and availability of the Invista, its most significant effect has been to boost confidence among customers who were wary of implementing the first version of a technically ambitious product

What’s one of the most significant differences between version 1.0 and version 2.0? The version number – 2.0, D’Errico said. Other differences include what EMC claims is a doubling of throughput, a five-fold increase in the number of concurrent data migration processes that the system can handle, and balancing of load across the back-end connections between the Invista and the storage arrays that it is controlling. One improvement made in response to customer feedback was to separate the two halves of the clustered out-of-band Control Path Cluster, allowing them to be physically separated as far apart as Fibre Channel connections will allow.

EMC says that migration is by far the most important application for Invista. 91% of customers in EMC’s Invista sales pipeline are interested in the product specifically as a means of moving or migrating data from one storage device to another, D’Errico said.

Hitachi and IBM have made similar comments about the usage of their competing TagmaStore and SVC virtualization products. That is especially so for Hitachi, whose virtualization software runs on its flagship TagmaStore disk array. Hitachi estimates that at any one time only just under 10% of the TagmaStore’s installed in Europe are using that virtualization to take control of any sort of other disk array. When it is not doing that, it is not delivering any of the advantages of heterogeneous replication or data migration, which are some of the major advantages associated with storage virtualization.

A lot of the use [of the TagmaStore virtualization software] is for transitory migration, or sweating assets for another three months before they drop off maintenance, Hitachi said.

Ironically D’Errico said that when customers ask about migration tools, EMC usually recommends products other than Invista, such as its array-based Open Replicator or SANCopy tools, or host-based Open Migrator software.

The field teams are comfortable with those, they’re simpler, and they fit with the arrays, he said. Invista is used when customers need more sophisticated migration tools that for example can provide one-to-many data movement, or a single point of control when moving data that is associated with one application but has been distributed over multiple disk arrays.

Our View

The storage virtualization market is contradictory (right down to the definition of what constitutes storage virtualization).

There is a whole host of potential reasons for the slow sales of the Invista. One of the most obvious is that it features a technically ambitious architecture involving exotic smart SAN switches, which many customers could find daunting. Another is that EMC itself might actually be a little half-hearted about selling a system that could one day free its customers from the need to buy expensive EMC array-based replication software.

But Invista’s slow going is not yet hugely important to EMC, as neither IBM nor Hitachi have made spectacular success with their competing products, and storage virtualization has not proven to be the market changer – or new source of customer control and lock-in – that it once threatened to be.