Hybrid cloud or half-hearted kludge?

Having lambasted Microsoft for its self-serving software-plus-services rhetoric the other week, today I'm going to apparently contradict myself and make a case for (some) hybrid clouds. Bear with me, though, there's method in this seeming madness. What irks me about Microsoft's position is that the company's spokespeople assume that any departure from a wholly cloud-based stance — any hint that there might be a few lines of software code running anywhere on the customer's premises — instantly validates their entire catalog of on-premise software. I find this immensely irritating because it seems to be an argument framed to justify their existing mindset, rather than attempting to engage with the new, cloud-centric model.

At the same time, I can understand that this mindset is prevalent not only at Microsoft but throughout the conventional software industry and among many of its customers. In fact, I think we all need to be much more aware of how pervasively the old, enterprise-centric way of doing things permeates all of our thinking and habits. I've been mulling this a lot recently in the course of preparing a keynote presentation for next week's Glue Conference in Denver: Gossamer and Glue: Weaving the Loosely Coupled Web, in which I'll be discussing the tension between our desire to open up to all the resources of the connected Web, balanced against our need to govern and protect our existing assets. A constantly recurring theme in the evolution of SOA, cloud and the Web has been the misplaced imposition of trusted, existing structures onto emergent patterns of interaction. This applies with special emphasis to hybrid clouds — build them to fit with your existing, unchanged infrastructure and you'll get little-to-no benefit. Change your enterprise to really leverage the cloud and nine times out of ten, you won't have any further use for a hybrid model.

Let me explain what I'm getting at here by reference to Cisco's recent announcement of an on-premise extension to the rebranded WebEx Collaboration Cloud (what WebEx used to call its Mediatone network before the Cisco acquisition). 'Ah-ha!' I can hear the Microsoft S+S evangelists crowing, 'yet another SaaS poster child acknowledges the need for on-premise components.' Look closely, though, and you'll see that Cisco isn't talking about replicating the entire WebEx cloud service on a totally independent on-premise server — which is what Microsoft means when it talks about a 'S+S' strategy for its own Web meetings platform. Instead, as Cisco's Alex Hadden-Boyd, director of marketing for the vendor's Collaboration Software Group, told me in a pre-announcement briefing, "You're extending the cloud into the premise but you're not moving off the cloud."

The new on-premises product is an add-on to a Cisco ASR 1000 Series edge router, and its purpose is to act as a local staging post when large numbers of users in a single network location are all accessing a WebEx session from the cloud. Let's say an organization has a training session from an external provider coming in to 1,000 employees in Chicago another thousand in London. Instead of a thousand individual sessions at given location each being served separately from the cloud, there's just one incoming session to the ASR, which then takes care of distributing those sessions locally. There's a huge benefit for the organization in reducing the bandwidth consumed as well as producing a better user experience, but the WebEx content is still coming from the cloud — and any users beyond the first thousand supported by the ASR can still connect directly via the cloud.

What we're looking at here, then, is a local unit that supplements a cloud service in the interests of a better user experience and more economic resource usage. It's a pragmatic response to the reality of having a large number of users at a single site, in which case you may as well extend the cloud to the site. It's the same principle as implementing Gears to offload part of an application's processing load to local clients (or, more ambitiously, Google's experiments with its Native Client). Why not move that part of the cloud service closer to the client if it makes the service faster and more scalable? So long as it's an optional feature rather than a requirement, it's fully consistent with a cloud philosophy.

What you might go on to wonder, though, is why this enterprise has so many employees on a single site anyway? That is what I mean by enterprise-centric mindsets. The Web is enabling distributed, remote working to an extent that is fundamentally changing the way enterprises are organized (organizations without boundaries, Cisco's CTO Padmasree Warrior called them in a recent series of blog posts on the future of collaboration). In a truly cloud-enabled world, such concentrations of workers may become largely unnecessary. This example illustrates why we should always be probing the motivations for pursuing hybrid cloud solutions. Of course, the world won't change overnight, and sometimes hybrid cloud is a sensible, pragmatic response to where we find ourselves. But all too often the desire to find some hybrid solution comes because we're thinking in terms of existing structures that perhaps we don't need any more. This is why I sometimes get impatient with people who look no further than the end of their nose and say they've embraced the cloud when they've barely begun the journey. The inevitable result of such thinking is they end up with some half-hearted kludge that's motivated more by a desire to avoid too much change and disruption than really seizing the opportunity presented by cloud services.