Going to the cloud: When on-premise DIY makes the most sense

Even if security concerns are put to rest, some companies balk at moving business functions to the cloud. Three big reasons: data archiving, application latency, and software that's not cloud-ready.
Written by Christine Parizo, Contributor

Image: iStock/wildpixel

Cloud computing isn't going anywhere. And for many organizations, moving applications, infrastructure, and other IT or business functions makes perfect sense. However, even as cloud computing matures, organizations still have legitimate concerns when it comes to letting go of their own in-house infrastructure, including data archiving, loss of control, and uptime.

IT leaders usually cite concerns over security when making the decision to keep their IT functions on-premise. However, if companies conduct due diligence and keep security risks and regulatory standards in mind when choosing a cloud provider, that can be a non-issue. Sometimes, experts will even say the cloud is more secure than on-premise deployments. So if cloud security is basically a non-starter, what other factors keep functions from being moved to the cloud?

SEE: Cloud computing policy template (Tech Pro Research)

Data archiving difficulties

Losing control of data archiving is one reason why companies may hesitate to move their operations to the cloud, according to Robert Douglas, founder and president of PlanetMagpie IT Consulting. "It's too easy to lose data in the course of the workweek, without active archiving taking place."

For example, a company might move business functions to Office 365. Then after a user leaves the company, their account could be deleted -- and along with that, email archives. In addition, deleted emails from multiple providers, including Office 365 and Gmail, may remain for only 30 days. Without active archiving, that information can be lost forever, Douglas said. To protect the company long-term, a separate archiving server on-premises or in a private cloud would be needed. And even if the cloud service does back up the data, there is always the question of where it resides.

Scale and latency of applications

"Working in the cloud inherently means that you're paying more per computer resource than you would be by purchasing hardware. The variable cost of managing and maintaining your operations is where cloud offerings shine," said Todd Millecam, CEO of SWYM Systems, Inc. However, some companies won't see cost savings by moving to the cloud. For example, midsize businesses that already need to pay IT staff to handle a lot of workstations, and large companies with ready access to DevOps talent don't need the management services from cloud providers, Millecam said.

"When your projected labor costs are less than double your hosting costs, cloud tends to not be a cost-effective option."

The latency tolerance of an application is also another consideration, Millecam said. Companies that are using low-latency applications, like scientific and medical applications, risk possible lag. "By moving to the cloud, you are sharing a computer with an unknown neighbor and are using an ever-changing network. That neighbor can be noisy -- or take up a lot of resources on the computer, and the path to get to your cloud server can get slightly longer unexpectedly."

While cloud providers do meet SLAs, it may take a couple of seconds for the cloud system to detect the latency issue. If a company can't handle a one- or two-second delay in applications every so often, the cloud may be a bad choice, Millecam said.

The end-user experience can be a huge barrier to cloud adoption, particularly for infrastructure-as-a-service (IaaS) or platform-as-a-service (PaaS) applications, according to Chris Sousa, vice president of solutions and strategy at Dataprise. "This experience is due to additional layers of complexity for accessing the application in the cloud," he said, citing remote desktop and VPNs as potential problems.

SEE: The Art of the Hybrid Cloud (ZDNet)

Applications that aren't cloud-ready

Lastly, some applications just may not be suited for the cloud, based on their architecture and support requirements, Sousa said. "If an organization has business-critical applications without an approved cloud version, migrating to the cloud may cause unnecessary risk and expense."

Organizations may also have built their own custom applications on-premise, and these would require extra time and effort to make them cloud-ready. "For custom applications, it puts additional strain on the development team that may not be equipped to handle management and support of these applications in the cloud," Sousa said.

In these cases, moving to the cloud may not make sense for companies. If the company doesn't want to give up control of data archiving, or if low latency is a big issue, staying on-premise could be the right approach. Some applications may not even be ready to move to the cloud, and re-architecting them might yield little in the way of ROI. For companies considering moving to the cloud, cost and security aren't the only issues to consider.

Also see...

Editorial standards