Cloud, and more specifically PaaS, is an interesting beast from an adoption point of view. It seems that everyone within enterprise IT loves PaaS, at least conceptually, but it is not yet the case that every new custom app being built by enterprise developers is targeting PaaS. Hesitation around PaaS adoption is often rooted in some variant of the “It ain’t runnin’ in my four walls, so it ain’t happenin’” mentality. Whether the rationale is founded in reality matters not; perceived hurdles might as well be real because they are real stumbling blocks to adoption.
Too many cloud vendors get caught up in trying to discredit a buyer’s rationale: “The public cloud is more secure” or “It’s a hassle to run stuff on your own, so you clearly don’t ‘get’ cloud.” Trying to change a buyer’s position on this requires herculean effort and you’ll likely come out bloodied and bruised, without a significant win.
If buyers interested in cloud and their cloud vendors step back and look for agreement in terms of cloud’s value as an operating architecture, they find a surprisingly long list of things that they do agree on. Things like increased agility, infrastructure flexibility, higher utilization and cost savings – all of which are unrelated to the outsourced service aspect of cloud. These values are fundamentally linked to a new IT pattern that is defined by operating and consuming a datacenter in a cloud-like way. This IT pattern (or operating architecture) is the same that cloud providers themselves use to drive cost down while sustaining quality. Almost by some sort of syllogistic axiom in rationalizing IT patterns, these virtues are ones that an enterprise would be keen on bringing in-house regardless of whether they adopt public cloud whole hog. It’s just a better way to supply and consume compute resources.
After all, if the cloud provider can use a new IT pattern to offer a service at low cost and high quality, why can’t a central IT department in a large enterprise do the same for its thousands of internal end users and developers? Many argue that these organizations just don’t “have the same economies of scale.” Maybe, but if we’re honest with ourselves we’ll also state that we know nothing about the slope of the curve that dictates economies of scale nor do we know how quickly marginal returns converge to zero. We also know very little in terms of what portion of an enterprise’s savings in moving to public cloud is due to the ‘outsourced’ nature vs. the tremendous agility gains (which are also present in private cloud). The idea that public cloud *might* be better should discourage any enterprise from pursuing alternatives that are demonstrably better than their current modus operandi.
PaaS in particular is interesting in this regard. PaaS defines a new way of thinking about infrastructure and applications, introducing new levels of fluidity and capability into how we build, deploy and run applications. PaaS, as a technology layer, defines application components and their fundamental interactions with one another as the unit-currency in IT. This is a stark change from an era where the first class citizen was always the infrastructure itself. By decoupling applications from infrastructure, a huge amount of agility and raw economic value is introduced (for example, through things like multi-tenancy). Developers benefit from (in a proper PaaS) having access to platform APIs and capabilities the equip them to build complex, modern applications. Everyone, developers, IT staff
So, should PaaS only be consumed in a public-cloud form factor? No! Essentially, anyone that takes this position is stating that, faced with a choice of using public PaaS only, deploying private PaaS, or doing neither, an enterprise should do neither for those enterprises that have true barriers to broad public cloud adoption. Having PaaS in-house means that development teams and IT pros can get new leverage out of existing assets, and experience order of magnitude efficiency increases across the board. I’m not suggesting that an enterprise shouldn’t adopt public PaaS (these types of posts always get misconstrued in this way); I’m suggesting that it needs to happen at the customers pace, and shouldn’t block adopting tremendous value in the near term. Enterprises will never rid themselves of their data centers in entirety. In fact, it is my opinion that enterprise datacenters will continue to grow at an alarming rate, and their consumption of cloud resources will explode.
By ensuring that an enterprise can leverage a cloud operating architecture in-house, it likely improves both the propensity to consume cloud resources in the future and shortens the window in which they will make the leap. An enterprise getting hooked to private cloud does the heavy lifting of aligning them with a cloud model, and later, their consumption of public cloud will go up. This “inside/out” model is likely the best bet to getting deep enterprise penetration, not the other way around.