Cloud computing services let CIOs reallocate up-front costs for hardware and software, but spreading such expenditures out over months and years will eventually catch up to your bottom line. And along the way, you may encounter many of the same challenges with data protection and content management that you would in your own data center.
Cloud computing costs creeping up over time to match and possibly surpass one-shot up-front costs is a reality that Dave Banks, chief technology officer of PropertyRoom.com Inc., says he's willing to live with.
Banks calculates that the Mission Viejo, Calif.-based Web business, which auctions off, stores and ships seized and abandoned properties for 1,800 police departments and municipalities, saved about a half million dollars up front by putting its new auction engine in cloud computing services provider Savvis Inc.'s data center. The savings encompassed servers, network infrastructure, software licenses and staff resources to configure the servers and network.
One significant up-front cost that was eliminated was the Oracle database license, which is included in the estimated $10,000 per month fee PropertyRoom.com pays to have its auction engine housed and maintained on Savvis' infrastructure. Within four or five years, the monthly fee will have caught up with the up-front costs Banks would have paid to house and maintain the auction engine in-house.
"Part of that tradeoff of saving up-front versus paying over time is that we were able to build another warehouse now, closer to the police departments we serve," Banks said. "That helps us save a lot in other costs like gas. Trucking goods across the country to warehouses is expensive." The company now has warehouses in Seattle, Chicago, Los Angeles, New York and Florida.
Backup adds to cost of cloud computing services
Savvis' monthly fee also covers PropertyRoom.com's backup and storage needs. The auction site's virtual machine images and database files reside on a shared storage area network (SAN) owned and maintained by Savvis, which also conducts backups that are stored locally and at an off-site facility for disaster recovery. The benefits of the shared storage model not only lie in up-front cost savings -- versus having to build and maintain a SAN -- but also in allowing virtual machines to be moved dynamically between physical servers in the case of hardware failures or capacity issues.
But data backup cloud services can create costs that sneak up on an organization. Chris Howard, an analyst at Midvale, Utah-based Burton Group Inc., compares the cost of usage-based backup cloud computing services with those of the phone companies for text messaging.
A client of his in Europe experimenting with the idea of going with cloud-based data warehousing via a pay-as-you-consume model soon found it would cost more to retrieve information out of that data warehouse than it would to keep the data warehouse in-house, he said.
"The cloud provider was charging for data access the same way cell phone companies charge for text messages," Howard said. "If you don't put the right usage terms in place from the get-go those [usage fees] add up pretty fast."
Cloud offerings have significant holes
Many organizations that do business on the Web are eyeing the cloud computing services model to save money but are not factoring in the cost of building in the same best practices they would use when building high-performance websites internally. CIOs need to check the limitations of the cloud providers' caching and self-provisioning capabilities, for example.
Depending on their cloud provider of choice, CIOs may still need a content delivery network, global load balancing and plenty of caching in their architecture, said James Staten, an analyst at Forrester Research Inc. in Cambridge, Mass.
Dave Bankschief technology officer, PropertyRoom.com Inc.
With the traditional hosting model, a call is made and the provider does all the work in the background to boost capacity. With the cloud, that capacity boost doesn't happen instantaneously either, like many people believe it does, he said. "It's up to your application development team to develop an interface that allows capacity to be turned off and on automatically," Staten said. "Don't assume that [cloud providers] are going to take care of all that for you."
Application development and testing in the cloud holds promise in that it cuts out a lot of the infrastructure costs to build a testing environment that mirrors the production environment. It also lets developers quickly spin up and down a test environment rather than taxing their own servers, or waiting in a queue to start a project in-house.
But the same rules apply for large-scale test and development projects, regardless of whether they're housed in someone else's massive data center.
"Testing and developing new applications is definitely cheaper on the cloud, and it's the best way to use the cloud on a short-term rental capacity basis," Staten said. "But if you are conducting large performance tests that simulate 200,000 people hitting that application, there is no cheaper way to handle that kind of load."
You may not need the cloud
If you are at a point when it is time to build a new application that requires a data center build-out, it is less expensive to tap a cloud computing services provider -- but have you done an analysis of untapped capacity in-house?
CIOs need to understand what untapped resources they have, not to mention when leases expire, before they go down the cloud services path. Burton Group analyst Anne Thomas Manes said she's seeing organizations that have several years left on floor space leases with hosting providers contemplating cloud initiatives.
"If you have excess capacity and leases that don't expire for another five years, it doesn't make sense to shift to the cloud because then you have all this extra expense and extra space that you have to continue to manage," she said.
Let us know what you think about the story; email: Christina Torode, Senior News Writer