Even since 3PAR’s inception, it has challenged the traditional assumption that a start up cannot deliver high end storage as it has steadily carved out a growing presence despite the odds stacked against it. Now it is challenging the new assumption that users are going to give preference to vertically integrated stacks over best-of-breed solutions. This defiant attitude was on full display as 3PAR’s CEO David Scott squared off with the analysts who were present at 3PAR’s first annual Analyst Day.
A blog entry I wrote back in November of last year came out mostly in favor of the idea of vendors providing vertically integrated stacks in virtualized environments from the server through the network down to the storage array. My sentiments were based on the premise that users are struggling with the complexity of heterogeneous server and storage environments and will grudgingly trade-off the extra cost and potential lock-in of vertically integrated stacks in exchange for their stability and predictability.
So I was somewhat surprised that, as 3PAR’s Scott began his morning keynote, he came out swinging against vertically integrated stacks. His comments obviously stuck in the craw of a number of the analysts in the audience.
They immediately started to fire back challenging Scott’s position. They claimed that the feedback that they were getting from customers indicated that customers wanted these new vertically integrated stacks and that best-of-breed environments were on the way out.
Scott did not back down and started to build his case for why best-of-breed would ultimately prevail against vertically integrated stacks. He argued that the industry is in the first decade of a 30 year transition to IT being delivered as a service from the cloud (private or public) and that the recent recession has only accelerated that transition.
He then contended that vertically integrated stacks have emerged from the predominant vendors (IBM, HP, EMC-Cisco-VMware) as a defensive measure against the cloud. The new cloud infrastructure threatens the hold that these vendors have on organizations so they had to come up with this vertically integrated stack concept as a means to slow – not accelerate – the adoption of clouds.
These vertically integrated stacks sound great in theory but pose two fundamental problems. First, there is the assumption that all of the components in the stack (server, storage and network) will work well together. That is a BIG assumption.
This might work well in the first release and maybe even the second. But to deliver on this architecture long term all of the pieces within the stack need to move in lock step.
That means that the entire stack can only move forward as fast as the slowest moving piece. So at some point one of the components in the stack will become the weakest link (unreliable, too slow, too expensive, etc.) and customers will once again revert back to a best-of-breed model.
Second, he asked the question, “Who is going to be the first provider in the coalition to give the customer a discount?” In the case of Cisco, EMC and VMware vBlock, it is a case of checkmate. The first one who flinches and gives a discount the one who will be expected to discount going forward. Further, they intuitively know that they have the customer locked into their vertically integrated stack so they have no competition.
It is for these two reasons that Scott believes that the vertically integrated stack is doomed to fail. While it might gain some momentum in the near-term, longer term it cannot stand up under customer pressure for faster innovation and increasing demands for cheaper, faster, better performing components.
He sees the early adopters of cloud (service providers and financial service) already foregoing the vertically integrated stack and adopting this best of breed model. He pointed to current 3PAR customers that include AT&T, Savvis, and Terremark (service providers) as well as Credit Suisse, NYSE and Regions Bank (financial services) that have looked at and already rejected the vertically integrated stacks because it did not give them the choices and flexibility they wanted.
So did Scott change my mind? I think he persuaded me that long term and possibly even near term best-of-breed will prevail over vertically integrated stack, especially in those companies with the technical expertise to deliver and support best-of-breed environments.
That said, 3PAR still needs to be careful not to cut off its nose to spite its face. Even though 3PAR may well be proven right, customers are hearing the message on many fronts that they should be looking for vendors that provide vertically integrated stacks.
This may put 3PAR in a position that if it does not provide such a solution, it may be perceived by some prospective customers as being behind the curve instead of in front of it. Therefore it may be in 3PAR’s best interest in the near term to investigate and potentially even offer a vertically integrated stack to satisfy those customers looking for such a solution until the market as a whole recognizes the value and inevitability of best-of-breed solutions.