What’s Happening in Cloudsourcing, Really?

Enterprises are coming to terms with the cloud, according to my most recent survey, but they are also recognizing that there are things that “classical wisdom” says they can do in the cloud that simply don’t work out.  This is going to change the risk/benefit profile for cloud computing for the providers because eventually the facts of cloud cost and benefit are going to get out.  In the near term, there’s an opportunity for cloud providers to step up and support the real world, getting a jump on competitors, and that creates a similar opportunity for vendors who can support cloud infrastructure in some way.

What everyone has ignored in cloud computing, say enterprises, is the cost of data storage.  The applications that meet business-case requirements so far are typically those that have relatively little storage required, hardly the mainstream of IT.  Most responsible studies of cloud adoption (in which, modestly, I’ll include my own) have always expected that penetration of cloud computing into the enterprise would be under 30% because of pricing.  What business buyers are learning is that basic virtual-machine hosting a la IaaS doesn’t address application support costs, and so that strategy cannot be expected to gain much savings from support economies either.  This more than any other factor is moving buyer consideration up the food chain toward PaaS and SaaS.

Still, according to enterprises, the big question is what will happen to data costs.  Cloud services exact a double penalty for cloud data; you pay for storage and then you pay for access.  Large databases that are churned regularly end up costing a mint in the cloud.  Transactional applications can’t deal with summary data (they create the data you’re hoping to summarize) so they are difficult or impossible to cloudsource.  IBM and other vendors have been pushing the use of summarized repository abstractions as the basis for BI applications in the cloud, and this notion is a good one because most BI doesn’t need to deal with detail-level data.  The result is that BI apps have outrun just about every other mission-critical app class; they dodge the data cost problem.

In the SMB space the picture is brighter in some regards.  SMBs are finding that the cloud’s biggest benefit is dodging technical support costs, which are higher for the SMB because of limited career path for good techs and often limited tech resources available in the local labor pool.  National recruiting is outside the scope of most SMB hiring.  SMBs also have literally an order of magnitude less data to store and move around, which means that their cloud costs are more easily covered by savings.  It’s hard to get a good estimate for SMB cloud potential in a realistic survey because SMBs are less likely to have a handle on either their current costs or their potential savings, but it looks to us like well over half of SMB IT spending could be cloudsourced based on our fall survey.  That says that the primary target for cloud services has to be the SMB…or even the consumer.

We’re starting to see some interesting strategies to try to make all this work better even at the enterprise level.  The most promising approach is to focus the cloud data market on backup or deeper storage, which means creating a transparent storage hierarchy that starts on prem and moves into the cloud, with applications running to move information among the tiers based on policy.  IBM, Oracle, and Microsoft are all said to be proposing this sort of thing, and a few startups have productized it in the form of storage appliances that virtualize such a hierarchy.  The tools aren’t perfect yet but progress is being made, and this may be the most important technology evolution in terms of maximizing the cloud opportunity.