Inside Advanced Scale Challenges|Friday, November 24, 2017
  • Subscribe to EnterpriseTech Weekly Updates: Subscribe by email

Avoid These Common Cloud Migration Pitfalls 

(ESB Professional/Shutterstock)

The vast majority of enterprises lack a mature strategy for at-scale cloud computing – and that can easily translate to inefficient or expensive cloud projects in 2017.

According to a recent survey by IDC, only 3 percent of 6,159 executives define their cloud strategies as “optimized,” the highest level of strategic maturity, while nearly half (47 percent) say their cloud strategies are usually “opportunistic or ad hoc.” Many allow project teams to choose and build their own cloud platforms, which can significantly accelerate cloud migration but lead to long-term cloud management problems.

After working with hundreds of enterprises to implement cloud technology – be it with a public cloud, in-house or hybrid – I have seen that the most successful enterprises do three things to get better returns on cloud investments:

  • Plan for cloud management, not just cloud migration
  • Start by focusing on a few platforms and tools to build internal competency (rather than jumping straight into multicloud or hybrid)
  • Implement proper account controls

The importance of a successful integration strategy

Gartner predicts that through 2019, every $1 that enterprises invest in innovation will require an additional $7 in core execution, a prediction that emphasizes the importance of a successful technology integration strategy. “Designing, implementing, integrating, operationalizing, and managing the ideated solution can be significantly more than the initial innovation costs,” Gartner reported. “Unfortunately, the deployment costs of the Mode 2 ‘ideated solution’ are not necessarily considered during ideation.”

Gartner’s prediction confirms Logicworks’ survey of 400 IT decision makers in 2016, which found that 80 percent of IT decision makers believe their organization’s leadership underestimates the time and cost required to maintain resources in the cloud. When we focus too much on the cost of the technology itself (the cost of Amazon Web Services compute resources, for example) or the cost of one-time migration, we lose focus on the more significant costs of developing a mature cloud operation team and processes to iterate off of early successes or failures.

Planning for long-term cloud management -- not just build-out or migration -- is the key to building an at-scale cloud computing practice.  Examples include:

  • Prioritizing automated, cloud-native tools that reduce ongoing management rather than prioritizing an “easy” migration (lift-and-shift of your existing tools into the cloud).
  • Prioritizing processes like writing resource templates (in a tool like AWS CloudFormation or Terraform), which require an upfront time investment, rather than asking your systems engineers to build one-off, snowflake cloud environments.
  • Investing in configuration management or working with a vendor or partner that already has a mature automation framework.

These decisions reduce the cost and time of maintaining your cloud environment. Even if you only partially automate your cloud, you still get benefits of treating your cloud like a piece of software: changes are documented and versioned, you can roll them back, and you (and your GRC team) have greater transparency into infrastructure configurations and how they evolve. You can create a common, shared library of templates and scripts so that multiple project teams can share learnings and reduce the effort of launching new projects.

Keep it simple

Conversations about hybrid and mutlicloud are on the rise. While these complex configurations make sense for many customers, companies new to cloud-based environments may find that building multiple clouds on Day 1 does more harm than good.

The most successful multicloud environments begin with an already-successful cloud operations team. The team becomes adept at a core set of standards on one platform. They learn their organization’s weaknesses and the limitations of that platform. Then they expand to more than one cloud if there is a very good reason to do so.

If you are on more than one cloud platform, you have felt the pain of scattered cloud management and probably see the value in standardizing cloud usage more consistently. The path forward is different for every organization, but in general: consider the option that is best for long-term management. Consider the simplest option -- usually one or a limited number of “approved” cloud platforms --, even if the easiest option is to let each team choose their own cloud. Organizations that do the hard work upfront to develop a single set of preferred cloud platforms with built-in standards for things like security and cost management have a much easier time expanding later on.

One cloud does not fit all, but if you pick a major IaaS cloud provider like AWS or Azure, one cloud certainly fits most.

Account controls

There is little hope for remaining on-time and on-budget without a feedback loop between IT and business. Business must receive recent data about the cost and performance of the project’s infrastructure, and IT must receive appropriate budgets in order to put thresholds and alerts around cloud spending. A simple set of account controls might look something like this:

  • Only engineers that directly work on the project have access, and only to the specific resources that they work with (Principle of Least Privilege). Example: Using AWS IAM Roles vs. granting the same access to every user or resource
  • Only certain engineers are authorized to create certain resources
  • All resources are tagged with appropriate project/team/developer so cloud bills are easily attributed
  • Higher-than-expected resource usage generates alerts
  • Certain resources have scheduled shutdown periods
  • Infrequently accessed data is periodically shipped to lower-cost cold storage (e.g. AWS Glacier)
  • Real-time costs are displayed in simple dashboard to IT and business (e.g. Cloudcheckr)

The granular cost data provided by these controls enable accurate planning for future cloud costs and, potentially, investment in something like AWS Reserved Instances (which allow you to save nearly 50 percent vs. on-demand compute costs).

It’s challenging to meet goals and achieve early ROI on your cloud investments. But whether you build a management framework for your in-house cloud platform or outsource it to a partner, it can make a substantial difference in your bottom-line costs and improve how business and IT work together – and even lay the foundation for cloud-enabled DevOps practices. Let’s hope that 2017 will not just be the year of increased enterprise cloud adoption, but also the year where our teams gain cloud operations maturity.
Jason McKay is senior vice president and chief technology officer at Logicworks.

Add a Comment

Share This