This morning, I listened to an interview Wikibon Analyst Dave Vellente did with Pat Gelsinger, EMC’s COO. Gelsinger and the EMC team thumped on Fusion-io today, casting it as a one-off, to gain attention for VFCache which it launched today with a big event in San Francisco.

During his press and analyst announcement, Gelsinger said EMC sold 24 petabytes of Flash last year. Fusion-io CEO David Flynn said to me in an interview today that his company sold 50 petabytes of flash last year. He said Fusion-io sold much of that Flash to the enterprise market, EMC’s customer base and scale out Web companies and cloud services.

That’s gamesmanship, which we love, of course. EMC is going to push aggressively on Fusion-io  to gain acceptance for VFCache in the PCIe flash card market. And you can bet that Fusion-io will be coming right back with its own retorts.

“EMC Project Lightning is the smarter terminal play, trying to distribute somewhat into the storage,” Flynn said. “Project Thunder is the mini-computer play. We’ve seen both of these plays before.”

That about says it. Fusion-io sees itself as the up and coming innovative pure Flash storage provider. Its bread and butter is replacing Storage Array Networks (SAN) with Flash. EMC is doing what we see a lot from established vendors. The company is augmenting its existing technology by adding VFCache and Project Thunder, its next generation flash caching service that will allow VFCache to be clustered in an appliance and distributed across a server network. This kind of challenge happens to ever successful company at some point. This just happens to exemplify EMC’s situation in its battle with Fusion-io.

On a much higher level, Fusion-io comes from an application standpoint. It admits the actual amount of I/Os it can pass is irrelevant to customers. It’s more about latency. That’s where customers suffer. The latency gap puts a big draw on applications. For EMC it’s about storage. That’s at the heart of the difference between the companies.

“We see a wholesale adoption of the distributed scale out model as we have seen with Web and cloud,” Flynn said.”The key thing is here whether it’s an open architecture or a closed architecture.”

The EMC architecture is not ready for the market, exemplified in the fact can’t support VMware vMotion.

“We had that capability more than a year ago,” Flynn said, explaining it as the reason it bought IO-Turbine last Fall. “EMC is so rushed that they are coming to market with something that I don’t think has a minimally viable capability set.”

EMC’s GeekFluent confirms in a blog post today that VFCache does not support vMotion.

What does this mean? A VM using VFCache cannot be moved via vMotion while VFCache is active. For the most part, since you’d only use VFCache for your highest-performing VMs, this should cause minimal inconvenience as you’d likely run those VMs on a separate compute-tier or use policies to ensure their access to resources.

Further, Flynn says, EMC does not have the controller technology it needs to keep its Flash costs down. It sources it to a third-party. As a result, Flynn estimates that EMC’s Flash is three to six times more expensive than Fusion-io. That’s why EMC only says its pricing is competitive with Fusion-io.

This is really a story with two different kinds of focus. Fusion-IO is application centric. EMC is seeking to add value to its existing storage arrays.

It is a race to the middle. EMC does need to maintain its margins and give a deeper way to sell its storage-based systems. Gelsinger fully admits that the world will go all Flash. But this is a time of transition. Customers are comfortable with the SAN systems they have but know that they are increasingly deficient to manage a new generation of applications. They trust EMC to deliver that mixed solution.

Fusion-io is the innovator. Their route is directly to the app. But they face the large task of continuing to show they can have the long-term reliability that EMC has developed through years of developing strong customer relationships.