By Minda Zetlin
The great thing about using the cloud is that you can scale practically to infinity. But that shouldn’t mean your costs for a cloud deployment are equally limitless. Too many companies, both large and small, start out with what seems like a very affordable cloud deployment, only to watch costs grow, little by little, until they reach unsustainable levels.
Here’s how to keep this from happening to your company.
1. Monitor cloud use. “You need service management and monitoring,” says Dave Goodman, director of digital strategy at technology research and advisory firm Information Services Group. If, for instance, you monitor a server hosting a particular application for 45 days, and usage never goes above 20% of capacity, you can decide to switch to a smaller server. Likewise, as an example, you can set an alert if you go over 70% utilization, allowing you to add capacity as needed.
2. Use optimization tools to get a handle on your entire company’s cloud use. “Cloud is very easy to consume and that can potentially be too much of a good thing,” says Chris Ortbals, executive vice president of product and marketing at data center provider QTS. In a typical large organization, he says, IT staff may provide or use cloud services, but people in Marketing and other departments might also have spun up their own cloud deployment, paying for it outside of the IT budget. It may not get deprovisioned when Marketing no longer needs it.
To address this problem, use a cost optimization tool to tell you whether you’re paying for servers or capacity your company isn’t using.
3. Don’t pay for more redundancy than you need. “There are oil companies calculating seismic data,” says Marty Puranik, founder of Atlantic .net. “They upload the calculator machines that they use, and then send the results to a high-availability database. They don’t really care if the machines doing the calculations crash because the data is in the database and they can just use another one.”
Those calculation machines are a perfect example of a cloud computing function that doesn’t need failover and disaster recovery systems. But they may get that protection anyway, due to cloud sprawl. “There can be a tendency for people in the business to start scaling things that don’t necessarily need to be scaled and turn it into billable hours,” Puranik says. “It’s very easy to have the costs get out of control with monitoring tools, failover tools, redundancy tools. It starts out as one project that then becomes much bigger and unmanageable.”
So make sure you know exactly how much you can spend, what kind of protection you need for which data, and what data and processes don’t need as much protection. Otherwise, Puranik cautions, you could wind up spending a huge sum to gain a fraction of a percentage performance improvement. “It can just keep going until the budget’s gone.”
Minda Zetlin is coauthor, with Bill Pfleging, of The Geek Gap: Why Business and Technology Professionals Don’t Understand Each Other and Why They Need Each Other to Survive (Prometheus Books, 2006).