When considering a move to the cloud, perhaps the most important question isn’t what vendor to use, but which apps will run best in the cloud. The answer will probably turn out to be some, but not all. Here are a few things to consider when thinking of clouding your applications.
First, there’s always the question of what sort of cloud you’re going to use—infrastructure, software, or platform? If your planned cloud platform is infrastructure/hardware as a service, then it’s a no-brainer. A pure IaaS vendor (Amazon’s close, as would one of the Hyper-V hosting offerings) only offers virtual machines and doesn’t give a hoot what software you run on it, so in that case software support is pretty much exactly the same as it was, pre-cloud—your current IT folks handle the support in essentially the same way that they currently do. You will neither incur extra costs nor cost savings on the support end, save for the ones I’ve enumerated in past articles (i.e., extra bandwidth and the like).
Next, is the cloud you’re contemplating one that’s built around a particular vendor’s applications, like Salesforce.com’s, a software-as-a-service (SaaS) sort of cloud? Well, now, that could be a winner or, at least in theory should be a winner. The whole underlying reason why anyone would even think of moving the lifeblood of their company—their organizational data—offsite is, as always in the cloud world, cost savings.
But is expecting that a cloud vendor can support an application more cheaply than our IT folks can a reasonable expectation? In certain circumstances it might be, for two reasons: returns to scale and better knowledge of the application. While it’s not always true that scaling up an economic operation lowers its per-unit costs (there are numerous examples of industries with constant or even decreasing returns to scale), the majority of IT-related enterprises should scale well. (In other words, setting up and maintaining one exchange server for one organization is expensive in terms of learning curve and/or salaries of Exchange experts, but once you’ve invested in putting in that first Exchange server, it shouldn’t usually cost anywhere near fifty times that amount to set up and maintain forty-nine more Exchange servers.) And by “better knowledge,” I’d hypothesize that if I did want to run those fifty servers and went shopping for a couple of Exchange administrators then I’d hope I’d get fairly good productivity if I could choose folks who actually worked for the Exchange team at Microsoft, or who had excellent access to that team.
If you want to see a cloud-based application support example where “bigger is cheaper and often better,” just look to what I’ve always considered the earliest set of cloud vendors: web hosting firms. The vast majority of websites are simple and the idea of an non-technical organization (or individual) spending the time and money to acquire a routable IP address, server class machine, web software, etc. to host its own website is clearly a mite nutty, and if we’re considering the range of applicability of cloud-based application support, web hosting for most website owners sits over on the “great match” side of things as a case of centralized expertise and economy of scale.
Exchange, Office, and SharePoint in the Cloud
Extending that notion, it seems to me that best candidates in theory for cloud-based application would be the most-used applications, and three obvious examples from the Windows universe immediately spring to mind: Exchange, Office, and SharePoint. So how’s that working out? Well, I hear mixed reviews on that score from friends and clients. Exchange hosting is the biggest success of the three by far, and I know a fair number of organizations large and small who have moved their email infrastructure out to the cloud either with an Exchange hosting service or Google’s cloud solution. In general the only complaints I hear are about downtime (and as I pointed out in an earlier piece, it’s essential to have some solid uptime stats on our existing networks before we can evaluate whether the inevitable and maddening downtime from a cloud vendor is really all that worse than what we were providing in the past) and the occasional synchronization problem. So long as you can get past the security and privacy issues associated with someone else’s computers holding all that confidential organizational message traffic, clouding email seems okay.
I don’t hear the same things about hosted Office and Microsoft’s shifts in the product seem to mean that Microsoft’s not so sure either. (Is it BPOS? Is itand, if so, will it still work next year on February 29, or what?) I don’t hear as much about customers running SharePoint in the cloud (and here I’m talking about SharePoint’s server-centric pieces, not the Office-centric stuff), but what I do hear is mixed between happiness and regret. My take on this, then, is that while there are cases where clouding apps makes sense, it doesn’t for most. (And don’t imagine that I’m happy to see that—given that I only need Photoshop for no more than about six photographs a year, I was really hoping to find a Photoshop-in-the-cloud vendor but, well, try Googling that to see what you get!)
Home-Grown Applications and the Cloud
I’m just about out of space, but I don’t want to leave this topic without briefly mentioning the class of applications on the other side of the range, applications that will almost certainly never make sense in the cloud—in-house, line-of-business applications. Spend any time with the IT staff of any Fortune 500 organization, and you’ll find that organization absolutely depends on a handful of home-grown applications, code cooked up by people who may not even work in the organization anymore and that is often written in obscure and old database languages, a bit of Cobol and a smattering of Office macros. As these apps have only one “customer” and are not publicly well-known (or known at all) and because those apps’ only experts work at the organization, the probability of a cloud vendor being able to support those apps even half as effectively as they’re currently being supported is right around zero.
Does that mean that those organizations should start moving from their in-house applications to more industry-standard apps? Some have suggested that to me, and when I told one friend that I’d recently visited a Fortune 500 that told me that they’d identified 317 essential in-house apps, she responded, “they ought to fire somebody,” but I can’t agree. When it comes to line-of-business apps, I’d much rather have something “hand-crafted,” so to speak, than something that’s generic and industry-standard. Take, for example, the case of banks. I’m simplifying, but basically all banks do the same things. So why are some banks far more successful, larger, profitable, etc. than others? For some successful banks, the “special sauce” that has enabled them to rise above their competition is a good set of line-of-business applications built over the years by IT folks who understand not only IT but the business that they’re in, and I think that’s true for most other sorts of large organizations as well. For many organizations, their home-grown, line-of-business apps are their secret weapon, and they’d be crazy to abandon that weapon, so I’d have to say no matter how successful the cloud industry gets, it won’t really find a home in support of line-of-business applications.
In sum, then, when considering moving things to the cloud, look to clouding the more generic software needs first. You’re most likely to find the low-hanging fruit there.