buchachon - Fotolia
Cloud repatriation is getting a good deal of attention and raising questions among storage managers and IT executives. Are companies that put data in the cloud really bringing it back on premises? Could it be that they're just taking time out to reassess their cloud storage strategies with respect to some data and some applications? Is it a big trend? Should you consider it? Or is it really just a few, well-publicized cases we're hearing about?
We asked some storage industry experts to weigh in on the cloud repatriation issue. A couple things are clear from their remarks: Cloud repatriation is real, but it's early days for cloud storage in general and any trend around it.
That was the perspective of IDC analyst Deepak Mohan. The repatriation of applications and data out of the cloud and back on premises is a reflection of the early stage of cloud adoption at many enterprises, he said in his email response. "Organizations are moving workloads to and from public cloud in the experimentation and familiarization process," Mohan said. "The overall momentum is centered toward public cloud."
Cost drives cloud repatriation
According the Taneja Group senior analysts Jeff Byrne and Jeff Kato, costs are behind any cloud repatriation that is happening. They expect cloud repatriation will increase over the next few years. "Customers are just beginning to see the issues around the cost of public cloud storage, the Taneja analysts said in their email response.
According to a recent Taneja Group survey, 60% of users are either mostly satisfied or highly satisfied with their cloud block storage. However, that leaves 40% who aren't all that satisfied and some may have enough issues to consider repatriation.
"Business issues are creeping in and leading some customers to consider moving their data back on-premises," Byrne and Kato said. "We believe that as cross-cloud data migration becomes less challenging, it will enable repatriation to be more viable than it is today."
The trend today, even when using a best-in-class hybrid cloud, is for customers to place their workloads in either a public or private cloud and keep them there indefinitely, Byrne and Kato said.
"Some innovative customers we have interviewed keep their source data on premises or in a co-location facility, push data to the public clouds for compute analytics and then retrieve only smaller subsets of results while abandoning or destroying the copy of original data they already have on-premises. This storage approach significantly reduces lock-in and avoids costly egress fees," they wrote.
Get ready for automated storage decision-making
Mike Matchett, principal consultant at Small World Big Data, said in his email response that cost is ultimately going to drive some companies to repatriate data back on premises. "The extremes of journeying all into the cloud or repatriating back out of the cloud will fall to the edges of most future storage architectures," and those are mainly going to be hybrid architectures, Matchett said.
These changes are happening fast, and they're going to be far reaching, Matchett added. "We are on the cusp of truly intelligent IT operations, in which we will soon see increasingly automated storage decision-making to the point where storage management itself may become transparent within five to 10 years," he said.