Mathias Rosenthal - Fotolia

Manage Learn to apply best practices and optimize your operations.

Learn from failure with a cloud project post-mortem

It's easy to point fingers at team members following a cloud project failure. But learning from what went wrong can help companies move beyond cloud disappointments.

Your cloud project failed. Now what? Look closely at your plans and learn from the results. Failures aren't fun, but they are important for moving forward with the cloud.

The first step in learning from a cloud project failure is to identify how the project failed. Almost all cloud failures fit into one of three categories: something didn't work, costs were too high or benefits were too low.

The success of any project, cloud or otherwise, depends on defending and meeting the business case with specific project steps, and this should uncover issues during the project rather than at the end.

The best way to start a cloud post-mortem is to meet with key project members. The meeting objective should be to improve the cloud project process, not set blame for problems. Avoid one-on-one discussions; everything should be out in the open.

Almost half of all cloud project failures aren't identified as problems until the cloud is complete and the cost/benefit validation has occurred.

Next, identify where the project's failure was first discussed. Almost half of all cloud project failures aren't identified as problems until the cloud is complete and the cost/benefit validation has occurred. This is a major problem because either there was no attempt to control costs and validate benefits during the project, or those steps were ineffective.

Identify snags before, during and after cloud projects  

Every cloud benefit that the team lists in a project plan should be associated with a set of assumptions; protecting those assumptions must be a specific goal. Be sure to identify each cost estimate and link to steps that can control costs, as needed. Miss this step and you won't notice the failure until the end -- when it's too late to try to fix issues.

If you recognize a problem while the project is running, find the point where it first occurred and ask three questions:

1. Why wasn't this issue brought up earlier in the planning phase?

2. Why didn't teams address this issue when it was first recognized?

3. Why did the project continue if the issue was not addressed and the failure was inevitable?

Tackle any issue that could result in failure during the cloud project planning phase. In most cases, a project problem is a planning problem. When planning a cloud project, address issues discovered and review how you created the planning process. Cloud plans should be "self-modernizing," meaning they reflect a learning curve. To meet the self-modernizing criteria, rethink your approach to cloud.

Immediately address any problems that arise after the project has started or halt the operation before problems get worse. If IT teams detect a problem and don't restart or end the project, review how they handled problems. Surprisingly, many cloud project plans lack specific procedures for handling issues.

Issue resolution 

An issue resolution plan is simple. Can the issue be resolved within the original benefit analysis framework? If so, take steps to fix the problem without affecting the parameters of the project's approval. If a fix takes you outside of the original benefit analysis, restart the project and re-analyze the business case. Otherwise, you could face other, bigger problems down the road.

If a problem poses a hazard to the business case, review the entire outlined cloud project plan. Do not attempt to shave costs or boost benefits within the initial project framework. The project doesn't necessarily need to be scrapped, but it should be put on hold until you can validate the approach. The most frequent response to a cloud project problem is to fine-tune things. While this may work, it's important to verify the approach itself isn't flawed.

All cloud project post-mortems could create more problems than they solve. Don't take on a process review to punish people's actions or decisions. Instead, work to define procedures that ensure future projects succeed or are rejected before they have a chance to become major issues. Don't look to blame teams; defensive mindsets kill projects. The cloud’s benefits are too compelling to risk, even to avoid potential failure.

About the author:
Tom Nolle is president of CIMI Corp., a strategic consulting firm specializing in telecommunications and data communications since 1982.


Next Steps

How to maximize the benefits of cloud

Private cloud projects start with hardware

How to avoid failure with your private cloud network

Dig Deeper on Managing cloud infrastructure