The cloud computing hype machine is in full throttle. Even some rational CIOs we talk with predict that 20% of companies won't own their IT systems in five years. From their perspective, what better way to get off the hook for high capital costs and poor system performance than to outsource? It's the ultimate in blame avoidance.
We're not buying that rapid change rate, but two things are clear from our annual InformationWeek Analytics State of Cloud Computing Survey and from the targeted Cloud ROI Survey we fielded in April. First, the movement toward adopting some level of cloud services is close to achieving critical mass. Second, there are gaps in how we're architecting, budgeting for, and engineering the services we buy.
Think your company isn't jumping in blindly? Don't be so sure.
Forty-three percent of our 607 respondents are either using or plan to use some type of cloud service within 12 months. That's up from 27% in our last survey. However, 71% say they have no idea whether that "cheap" cloud service will end up costing a fortune. They're making cloud decisions in a vacuum, bypassing the review that would, as a matter of course, happen for an internal IT initiative. Yes, we're talking about the checklist gauntlet: How much does it cost up front? How much to maintain? How does it connect? How do we monitor it? How do we back it up? How do we secure it? How does it impact everything else? What do we do if it goes down?
"The driving factor for many cloud decisions is cash flow," says Jeff Solomon, head of the emerging technology group at CPA firm Levine, Katz, and Solomon. "For many industries, the capital limitations are so dire they're willing to move to the cloud based on a cash flow comparison alone. They simply aren't doing a proper ROI analysis."
The up-front cost differences between cloud and in-house are so dramatic on rudimentary analysis that businesses are lulled into feeling savings are inevitable; it's the "truthiness" factor, if you will.
As we discuss in more depth in our full 2011 State of Cloud Computing report, it's not that we don't like the cloud; in many cases, providers can deliver a superior-quality service for less money. That said, you don't have to look back too far to see an example of how poor financial justifications hamper long-term initiatives. A case in point is virtualization. The 2006 pitches to CFOs were based on a pure hardware vs. virtual platform analysis and left out the cost of staff retraining and systems management software. And let's be honest: Those omissions weren't just because neither area was fully developed yet. Including them would have weakened the savings argument.
Flash forward to 2011, and many companies are struggling to manage complex mixed physical and virtual data centers because they lack the funds to get the management tools and staff training they need. Integrating cloud services into your enterprise has an even more knotty set of challenges. Simply comparing the recurring monthly operating expenditures vs. capital expenditures with some net present value thrown in for good measure will bite you in the long run.
Crunch The Numbers
Before approving a plan to run any IT service in the cloud, there are 11 areas where you need to calculate costs. We delve into those more in our customizable spreadsheet, but start with a five-year baseline of comparison. Yes, this may seem like a long time frame--especially when considering hardware costs--but we think it's fair. Not only does it give a more accurate representation of the true average life of hardware and software, but it also forces you to address longer-term investments in bandwidth, monitoring, and failover. Every comparison needs to include the following:
> Hardware: Include router or WAN appliance upgrades vs. server or storage hardware purchases.
> Software: Not usually a factor on the cloud side of the ledger, but internally, include upgrade costs.
> Recurring licensing and maintenance: Typically monthly for cloud, yearly internally.
> Bandwidth: In the cloud column, assess a portion of your current Internet charges plus planned upgrades.
> Staffing allocation: Even cloud-centric shops require some staff for ongoing administration and engineering. It's likely to be a rough estimation, but include operations and application development allocations.
> Monitoring: In-house apps typically get a no-additional-chargeback status, if you've invested in enterprise monitoring. Cloud initiatives probably will require an expansion of your monitoring system, most likely adding Web triggers or purchasing a third-party cloud app.
> Backup and archiving: A major miss for many--you need a plan for backing up and archiving data stored in the cloud. This may end up being an additional cloud provider for storage, but it needs to be included.
> Failover and redundancy: Your business continuity plan should have an allocation formula for various systems. If the service becomes unavailable, what is your alternative? Figure it out and budget for it. Period.
> Security audit and compliance: Budget time and resources for adding the cloud initiative to the compliance officer's stack of auditable items. Make no mistake: Sensitive data in the cloud is still your responsibility.
> Integration: This is especially critical for software-as-a-service and core database initiatives. Calculate development, maintenance, and licensing as separate line items.
> Training: Yes, you need to factor in training, regardless of whether you're adding software, infrastructure, or platform as a service. The claim that a service is "self-explanatory" is almost always bogus.
Finally, we suggest you add one other factor into the equation: speed to implementation. What's it worth to be able to add and integrate a project management system in 120 days, not 220? Is there a competitive value you can place on off-loading batch processing jobs to the cloud in three weeks? Chances are you'll end up with an arbitrary number, but the exercise forces the business--not IT--to place a dollar value on speed.
Even if your five-year numbers favor in-house, you may still opt for the cloud due to cash flow constraints. But at least you'll have the longer-term costs laid out for the CFO. Even if cash flow isn't an issue, there's still a benefit: licensing discounts from cloud vendors of up to 30% for paying on an annual basis. "The entire tech industry is transitioning to or adding a cloud-based revenue model," says Solomon, from Levine, Katz, and Solomon. "They need cash, too."