Many CIOs have their doubts. In this series, we examine why cloud costs remain so hard to judge.
What's the answer to skeptics who think falling hardware prices aren't matched by falling cloud prices?
One answer: Public cloud infrastructure as a service, such as Amazon Web Services EC2 or CenturyLink Savvis, is a system and not just a hardware component, and the prices on systems have never dropped as steeply as components. The buyers of disks or CPUs that keep up with Moore's Law still have to install, operate, and update those systems themselves, while the cloud user can fire the people who were running the on-premises IT systems or free them up to take on new business development tasks.
However, rarely do we hear the cloud business case made around the cash savings from firing a lot of staff--just 21% of the respondents to our IT spending survey expect reduced head count long term.
Anecdotally, the business case is often made around agility--dealing better with spikes of activity or moving more quickly to meet a changing business need. That's the business case that worked for GSD&M, a 500-person Austin, Texas, advertising agency, when it pitched an online promotion last holiday season for Marshalls stores.
Late in the 2011 holiday planning season, GSD&M came up with a "Share a Carol" promotion. The agency would record carols using four actors, then let Marshalls' customers online put them in front of a Google Maps Street View image of a friend's house--or most anywhere--and send a link.
Marshalls' marketing department liked the idea, so GSD&M scrambled to find developers to produce the Web app. But Marshalls' marketing department didn't have the computing capacity on which to run the app; provisioning servers in the company's data center would take time. So GSD&M turned to the Rackspace Cloud, which provided the servers to quickly launch the program and could also scale up if traffic spiked beyond projections.
As those projections rose 50% just before the project was launched post-Thanksgiving, GSD&M told Rackspace that it needed to double the server memory and storage. "In the cloud, we were able to change those resources on demand, as the concept changed," says Jerry Rios, GSD&M's CTO and senior VP of information. Getting that kind of variable capacity from Marshalls' IT organization would've run into "complications," he says.
Rios won't say what he paid for the Rackspace servers, but he says Marshalls' marketing could pay for it out of its own budget. GSD&M went from a small staging server during development to doubling the memory and storage of a production server when estimates rose, and then also extended the campaign six weeks because of its popularity. Without that ability to ramp up the IT infrastructure quickly, the promotion wouldn't have come off, Rios says.Price isn't always the top criteria when you need to move fast.
Another cloud infrastructure user, Broken Bulb Game Studios, is doing a different calculation--of dedicated cloud-based servers versus multitenant ones. The online game builder hosts 371,000 daily players of games such as Braaains and Miscrits of Volcano Island on 20 dedicated servers and 10 multitenant cloud servers, both managed by service provider SoftLayer. Broken Bulb is now moving all of its games to multitenant servers because it finds they provide the same reliability and manageability as dedicated servers at a lower cost.
"We are not venture-capital-backed. We can't tack on an extra $10,000 server whenever we want one," says CEO Robert Nelson. From its startup, Broken Bulb avoided building its own data center to concentrate on producing game code. It's moving 60 TB of data from dedicated servers to multitenant servers.
Broken Bulb uses RightScale as the front-end manager of those new cloud instances. RightScale has monitoring and analytics, so it predicts when a virtual server is running out of capacity and prepares another. RightScale services typically add 3 cents an hour to a service provider's regular charges.
One reason Broken Bulb is saving money is because it cut in half the RAM and processing power of its multitenant cloud servers compared with its dedicated servers, while still getting all the performance it needed. Nelson isn't sure whether SoftLayer manages its multitenant servers more efficiently than it could its dedicated servers, or if Broken Bulb had simply over-provisioned itself.
Coupa Software is typical of how cloud infrastructure has changed the way startups launch. The maker of procurement software got off the ground in 2006 focused on the system it wished to build rather than building out its own data center, using software-as-a-service and public cloud infrastructure services.
But it has also become a more savvy user of cloud services, cutting its cloud computing expenses almost in half over the past couple of years, says Sanket Naik, Coupa's senior director of cloud operations. To get those savings, it adopted Amazon's reserved instances--a commitment to a year of use plus an up-front payment in exchange for a lower hourly rate.
Naik wants Coupa to invest in technical people to build its core product, not to uncrate servers to be racked and stacked. And he thinks the price of computing in the cloud is going down, citing a March 6 price reduction from Amazon.
Each of these companies has a different yardstick for measuring the value of cloud computing infrastructure, one that doesn't hinge entirely on comparing the cost of public cloud services with the cost of on-premises computing. What they have in common is that each company would resist a you-can-buy-it-and-run-it-for-less argument. Cloud brings them an efficiency they don't think they could replicate in-house, and it lets their tech pros focus on new business projects over IT operations.
Multicloud Infrastructure & Application ManagementEnterprise cloud adoption has evolved to the point where hybrid public/private cloud designs and use of multiple providers is common. Who among us has mastered provisioning resources in different clouds; allocating the right resources to each application; assigning applications to the "best" cloud provider based on performance or reliability requirements.
InformationWeek Must Reads Oct. 21, 2014InformationWeek's new Must Reads is a compendium of our best recent coverage of digital strategy. Learn why you should learn to embrace DevOps, how to avoid roadblocks for digital projects, what the five steps to API management are, and more.