The realist’s guide to cloud services and what they’re good for

Hot technology chatter, like most other business discussions, assumes business success comes from brilliant concepts, careful planning, and disciplined execution. I wonder if what actually happens is more of the million-monkeys-at-a-million-keyboards situation: If enough companies throw enough products and services into the marketplace, some will stick, even if the products and services are the result of random tossing, not superior thinking.

Way back in 1999 I proposed a simple set of criteria to bring order to the randomness when it comes to new information technology. According to the proposed model, three filters predict the success or failure of any new technology product:

1. Are the customer and consumer the same people? For clarity: “Customers” are those who make or influence the buying decision. Consumers are those who actually use the product or service.
2. Is the product or service affordable? The third “customer” role is the wallet — the person or committee who approves spending. Products are affordable to the extent the wallet doesn’t care about the cost increment. The wallet impact is a separate factor whether or not the wallet is also the customer or consumer.
3. How much disruption it causes in the enterprise it’s brought into, as opposed to “the marketplace,” which doesn’t matter a bit.

So here we are, with a new year in front of us, after lots of hype about the three big IT trends: The cloud, BYOD (bring your own device), and big data. It’s past time to apply this model to them and see where it takes us. This week we’ll see how the cloud stacks up. We’ll look at BYOD and big data next week.

Assessing cloud success

The cloud, just to make sure we’re on the same page, is any technology that’s bundled into services and made available through the Internet. SaaS, PaaS, and IaaS (software, platform, and infrastructure as a service, respectively) are the most important. One at a time:

Software as a service SaaS has been tremendously successful, assuming you think press releases and exponential growth forecasts constitute success. Hard data on SaaS market penetration by software category is remarkably difficult to find — and that’s what matters. In case the point isn’t clear, Salesforce.com has penetrated the CRM/SFA (sales force automation) category quite successfully. TeamDynamix has done well in the project management category.

But ERP, supply chain management, warehouse management, point-of-sale, business intelligence, or even basics like office suites? Extravagant growth projections are easy to find. On the other hand, accurate reports of current market share are not. In any event, here’s the analysis for SaaS, according to our three criteria:

Customer vs. consumer: Business executives and managers are the customer. Business managers and staff are the consumers. Organizationally, they’re the same.

Affordability: For SaaS solutions, the customer is the wallet. For installed software, the capex committee is the wallet. This makes SaaS much more politically affordable, even though it’s generally less economical than its installation-based competition.

SaaS flies in under the CFO’s radar because buying SaaS licenses doesn’t involve the capital approval process — they arrive divided into 12 hard-to-spot installments per year. Installed software, in contrast, requires capex funding. It shows up in one big bite that requires capex committee approval.

SaaS isn’t actually economical, though. Take the ever-popular Salesforce.com. The last I looked, licenses cost $780 per year, per user. Compare that to Act Pro’s one-time $550 per-license cost, and it’s clear Salesforce.com actually costs more — a lot more.

But that doesn’t matter. When dealing with the CFO and capex committee, maintaining ignorance (or plausible deniability) is blissful.

Disruption: It’s all about integration with other systems. Salesforce.com, the highest-profile SaaS success story, is usually an “automation island.” Most software categories require extensive integration, though, and SaaS integration is significantly more complicated (and consequently disruptive) than what IT has to do to with software installed in the data center.

Assessment: For SaaS products with high up-front integration needs, the outcome is questionable. Beyond SaaS being disruptive, once IT has to get involved, the CFO is more likely to look at the overall economics of the decision.

SaaS is and will be a winner for applications with limited initial integration requirements. Once a package is entrenched, IT will almost certainly have to integrate it later on, but that doesn’t matter.

Platform as a service PaaS has been the cloud’s orphan stepchild, but that seems to be changing. Assembling, maintaining and administering a modern n-tier development/test/production environment is, shall we say, a nontrivial task — unless IT signs up with a PaaS vendor, at which point it becomes someone else’s problem.

Customer vs. consumer: The CIO and IT management are the customer. App dev and IT ops are the consumers. Organizationally, they’re the same.

Affordability: Political affordability doesn’t end up in PaaS’s favor. PaaS is opex, making the CIO the wallet. CIOs often prefer capex, because that makes the capex committee the wallet — the cost no longer comes out of the IT budget, which is attractive to your average CIO.

As for PaaS’s actual affordability, the financial analysis probably depends on how the person doing the analysis wants things to come out, and solving for the preferred answer won’t even require much ingenuity.

Disruption: Right now, PaaS is disruptive in two or three respects, depending on which vendor you choose:

• Extending management tools to your PaaS environment isn’t yet a solved problem, especially with respect to identity management tools like Active Directory.
• Integrating PaaS-housed data and business logic with existing production systems is even messier than when everything is inside the data center — no messier than with SaaS, but in the case of PaaS, where IT is the customer, integration is what’s expected.
• Many PaaS vendors offer proprietary development environments. That’s disruptive for reasons too obvious to mention.

Assessment: PaaS is, sadly, a dicey proposition. It’s politically less affordable than owned platforms and sufficiently disruptive that, except for capital-starved startups, it has a steep hill to climb.

Infrastructure as a service IaaS is the least interesting cloud variant. It’s where IT can rent virtual hardware, so instead of buying enough capacity to handle peak processing loads, IT can rent only as much capacity as it needs right now — handy when processing loads vary seasonally or are unpredictable and spikey.

Customer vs. consumer: As with PaaS, the CIO, IT management, and app dev and IT ops.

Affordability: See PaaS. The politics go one way, the financial analysis might or might not go the other way, depending on who has control of the spreadsheet.

Disruption: Take the first two bullets listed for PaaS. Add the absolute requirement that, by contract, each IaaS vendor must guarantee all storage must be provisioned in the same data center as its processors, linked at wire speeds. You really don’t want database accesses to cross the continent at MPLS or slower data rates and latencies.

Analysis: For startups planning to use IaaS as the most modern approach to data center outsourcing, it is (or can be) a winner. For stand-alone applications that need no integration and limited management, IaaS can also be a winner, especially when processing loads aren’t predictable. As an extension of an extensive installed base of applications running inside a preexisting data center? Questionable.

Taking it home The problem with overhyped technologies like the cloud is that they’re touted as panaceas that will solve all of the computing world’s problems, which means those who are skeptical must be chumps.

This chump predicts that the cloud will have a place. It’s a much smaller place than what you’ve been reading, but a place nonetheless.