"The early bird gets the worm, but the second mouse gets the cheese."
As a bona fide technology geek, I have to confess that the success of Apple Computer over the last decade was hard to comprehend at first. Think about it from the perspective of a guy who was always at the bleeding edge of gadgetry: a digital media player with a proprietary file format? Seriously? I owned several of those. The kind that used “open” MP3 files. A smart phone? Had one of those already. For years, actually. A tablet? I worked for the company that sold the Palm Pilot in the 90’s. Old news. It wasn’t that I lacked admiration for the actual computers and laptops. In fact, I admired them for their simplicity and capability. It was just that I didn’t get the strategy of trying to be a follower in areas like this.
The mental hurdle I had to overcome was the same one that technical people everywhere seem to stumble over: the assumption that everyone is technically inclined like me. Once past that obvious fact, it only took a little bit of imagination to understand the success of these products. It was all really attributable to seamless integration of the entire end-to-end experience. The ability to easily buy, store, and play music was far more important than the file format. Extending that ability further with a phone that is always connected and runs software was better than the little web browser on my Nokia. Applying that same experience to other media in a larger form factor superseded the need for a stylus and handwriting recognition. It turns out that much money can be made selling technology to people who want to use it without having to understand it.
A similar evolution has been taking place in enterprise IT. Enterprise data centers are arguably the pinnacle of technical complexity. Within them, IT teams are tasked with piecing together useful business process automation by assembling parts from a Lego bin composed of network, compute, storage, databases, development tools, and consulting services. For the past several years, the vendor trend has been to pre-integrate some of these pieces. Back at Equallogic, our success was attributable to the product’s high level of automation and integration, which allowed administrators to stop worrying about piecing together working storage systems from disparate software and hardware. Instead, they could just buy integrated, intelligent appliances. The same all-in-one strategy has proven to be valid in networking, data warehousing, and even system management. The more recent success of hyper-convergence – collapsing storage, compute, and networking functions into a single box – should surprise no one. Time and again, it has been proven that IT prefers to focus on real business problems rather than the vicissitudes of building infrastructure.
Although integrated appliances are clearly superior to cobbled solutions made from disparate parts, convergence and hyper-convergence are not going to be enough. Nor are unholy alliances between infrastructure vendors to build “solutions” (think VCE). In the super-complex environment where large enterprises live, even the simplest task becomes an epic time sink. Integrated appliances can only make minor improvements when technology needs to be deployed at massive scale. The forklift upgrades and never ending equipment acquisition cycles are a frustration to everyone. Policies, processes, accounting, chargeback, and organizational dysfunction all conspire to slow down innovation. Most organizations must bear the cost of extensive professional services just to make technology consumable by their employees.
In order for these problems to be solved, the technology stack will be rehashed and re-integrated much further than we have seen up until now. Regardless of whether it’s from the big three cloud vendors or from a new group of upstarts, the next generation of enterprise infrastructure will look more like typical voice or cable TV services: minimal equipment, a wire, and self-serve provisioning of limitless resources. Like all the iWidgets, it will be made up of largely conventional components that are hidden from view. Nevertheless, it will succeed because it is far easier to work with. It is still very early in this technology cycle, but a decade from now, infrastructure will be universally consumed in a pay-as-you-go fashion. The ultimate form of infrastructure stack integration is not a box, it’s a service.