Why Your Cloud Migration Strategy Is an "F"

It is an open secret and the subject of an increasing number of discussions. There is growing dissatisfaction with cloud migrations, and some organizations are increasingly discussing repatriating workloads back on-premises.

According to Gartner’s Dennis Smith, the problem lies not with the cloud itself but in how organizations approach their cloud implementations.

Many organizations, he says, might think they have a working cloud strategy, but in reality, they don’t, or they might lack one altogether.

This could be because they consider IT needs without input from other parts of the organization and operate under the false assumption that a cloud strategy is the same as a cloud implementation plan or a data center strategy.

Smith, a distinguished vice president and analyst in Gartner’s Infrastructure and Operations organization, expanded on these issues in a recent presentation at a conference in Sydney.

It’s the business strategy, stupid!

While acknowledging that having a cloud strategy was a “business necessity,” Smith outlined the main challenges he had observed through thousands of interactions on the issue.

“The first mistake is assuming that it is an IT strategy only and doesn’t involve the other elements,” he told the Gartner Infrastructure, Operations and Cloud Strategies Conference in May.

“When you look at a cloud strategy, it has to involve a business strategy that links to the cloud strategy and the operational strategy. Oftentimes I find myself reviewing strategies that are 150 pages, but it’s not a cloud strategy. A cloud strategy is concise, it is very much to the point and really addresses the ‘why’ of implementing.”

Smith said he had often spoken with clients who had gone through the “muck and ire” of implementing a cloud strategy, only to find that they don’t have a cogent answer as to “why’ they had done it from the point of view of the business.

Exit strategy

In some cases, cloud providers experienced significant outages, and clients realized they were not being provided with the capabilities and features they initially believed they had.

In other cases, there were competitive intersections with the cloud provider, which raised questions about the relationship's viability going forward.

“In these cases, it is vitally important to think through the exit strategy very early, and it doesn’t involve, in many cases, SLAs so much as technology architecture,” said Smith.

“If you are tethered to a cloud provider vertically and using all the capabilities of that cloud provider, it is very difficult to exit from that because your developers, workflows and pipelines are associated with that vendor. So the push and pull of going vertical versus commoditizing that vendor’s capability has to be factored into your cloud strategy.”

“Oftentimes, I find myself reviewing strategies that are 150 pages, but it’s not a cloud strategy.”

Cloud strategies also need to be flexible and re-evaluated every 12 to 18 months to ensure that “it is still keeping up with your actual requirements.” This makes the cloud strategy a “living document” that changes with the business.

“Let’s say your executives decide that now they are looking at acquiring companies, so you are being forced into a multicloud strategy,” said Smith.

“In that case, you may not be able to dictate that this is our preferred cloud provider, but you will now have to implement a blueprint around bringing on these other entities, so it’s good to keep these elements in mind so you are never static.”

Beware “newly minted” executives

Another frequent mistake Smith has observed is when a “newly minted” executive joins the business demanding to “move everything to the cloud in 18 months.”

“This isn’t a strategy in many cases; it’s a misguided statement,” he said.

A better approach is to take an “application by application” approach without “oversteering” and being too aggressive with the migration.

It is advisable to be “more pessimistic” about what cloud computing can actually achieve.

“Once it’s all said and done, you probably won’t have everything in the cloud, and most organizations I talk to will still be hybrid with a meaningful workload in an on-prem environment,” said Smith.

IT leaders should be concerned with executive mandates on major cloud migrations because it often means that workloads that move to the cloud might have to be moved back again due to lacking due diligence in the decision-making process.

Venn diagrams

Another mistake is taking the “all or nothing” approach with the data center strategy and equating the data center with the cloud strategy.

“Think of it as a Venn diagram, where the cloud strategy is in the middle, and there is a separate data center strategy,” said Smith.

“There may be a strategy for security, one for application development and so forth. The intersection of your cloud and data center strategy will be heavy, particularly if you’re doing any AI activity because many organizations will do much of their AI activity within a public cloud. But it means taking data from on-prem and into the data center environment.”

Smith also addressed the issue of cost, another key source of dissatisfaction, because the reality of the consumption-based model is often underestimated.

“It’s like owning your car versus taking an Uber for everything, and there’s a very different level of governance,” he said.

Perceived advantages in cloud migration — agility, speed and the ability to automate — were often not being delivered upon.

“This can be dangerous, particularly with a lot of the vendor changes and vendor activity which is going on out there,” said Smith.

“I’m not suggesting that you should be vendor-neutral, but I caution against being totally tethered to a vendor because if that is the case, you will…not have an exit strategy.”

Image credit: iStockphoto/IcemanJ