Cloud computing may have been a good idea at the time, but it isn't always right for your application. How do you move it back to your own systems? Experts offer advice.
Suppose that your company decides to leave cloud computing behind and bring everything back into its own sunny data center—go ahead but proceed cautiously, experts say.
There's no particular trend away from cloud computing and toward insourcing, but there are valid reasons why organizations may make the change, which include:
- Maybe your business needs are reduced;
- The cost savings aren't what the service provider promised;
- There's too much downtime;
- The latency is too slow for your application;
- Your budget changed from being operations-centric to capital-centric;
- A security concern;
- A corporate merger/acquisition where the bigger company has different plans;
- There are new in-house expert employees; or
- It's simply a matter of control.
Thus, the issue of how is more important than the question of why.
SEE: Cloud migration decision tool (Tech Pro Research)
There are plenty of things to do and to not do, said David-Kenneth Group's Kris Mathisen, senior vice president of the Maryland-based IT consultancy. Whether it's a cloud provider or their own operations, companies tend not to be honest with themselves about organizational maturity, skill sets, and infrastructure, he said.
When planning a move back to on-premise systems, "You have to understand the relationships between the applications, the infrastructure, the application interfaces, and the business environment," Mathisen noted, adding that such concerns can become exponentially worse when applications are based on microservices or distributed across wide-area networks.
"When you start breaking those things apart, you can start breaking those systems. You have to understand those things," Mathisen said. He reminds clients to keep in mind that some system components may synchronize hourly or weekly or monthly or quarterly or annually, so what you inadvertently break now may not be obvious until much later.
"You've got to understand the interoperability and data portability," Mathisen continued. Software-as-a-service companies never admit this, but they love vendor lock-in. "It's kind of like the roach motel. You know what they say, you can check in but you can't check out," he joked.
The people matter, too. For example, your IT department may have shifted its personnel away from system administration experts and toward network management experts when things got cloudy, but if you move back, you need to change the personnel yet again.
Finally, Mathisen said, "You need to make sure you re-align your service continuity and your disaster recovery plan so it represents the new paradigm that you're getting into." It would be easy to overlook long-ago plans for disaster recovery, which now might be obsolete after bringing applications in-house, he explained.
SEE: Disaster recovery and business continuity plan (Tech Pro Research)
Speaking of disaster recovery, there's software made for that purpose from many vendors, and it could reasonably be re-purposed to help companies leave a cloud. Veritas product manager Vidhi Kastuar made that point in a conversation about the company's CloudMobility software. The application launched in 2017 with a focus on helping companies move from one cloud to another. Features for moving from a cloud back to your own location will be part of a significant update due in the next couple of quarters, she said.
Kastuar suggested that if companies are using public clouds and wish to bring their data back home, then they may want to evaluate private cloud options as a less complicated change while still gaining some control of their systems. It doesn't have to be all-or-nothing, she noted.
4D Data Centres, located in the UK, provided additional comments from Operations Director Steve Wright. "The number one discussion we have with CIOs is around costs and cost inflation. While on the face of it, the elasticity of public cloud providers gives great flexibility, the majority of services delivered for the enterprise have relatively predictable workloads," he said. "The additional cost for flexibility is negated by this predictability. Enterprises are also caught out regularly by bandwidth charges either within the cloud environment or into their offices. Services such as direct connect or peering can help to reduce this, but this lack of control and precision has been a significant cause of tension between the CIO and CFO."
"In approaching the plan to de-cloud parts of the enterprises IT services, it is critical to understand the relationships between systems and proprietary services that may have been built within the cloud service provider's ecosystem. However, the number one starting point for this assessment is at the infrastructure-as-a-service areas. These are generally the fastest and easiest to move and have fewer direct tie-ins restricting movement away from the cloud provider," Wright continued. "Significant cost efficiencies can be made by looking into these areas of predictable resource requirements and looking at the total cost of ownership over either 24 or 36 months justifying the potential capital expenditure required."
- Everything as a Service: Why companies are making the switch to SaaS, IaaS, PaaS, and more (Tech Pro Research)
- Cloud computing: Three reasons why it could be time to go cloud-first (ZDNet)
- Going to the cloud: When on-premise DIY makes the most sense (ZDNet)
- 62% of enterprise IT leaders say on-premises security is better than cloud (TechRepublic)
- True private cloud isn't dead: Here are the companies leading the charge (TechRepublic)