Performance Management for Complex Multi-Cloud Deploymentsv1.0

1. Summary

As enterprises migrate as fast as they can to the cloud, the increased complexity of cloud computing deployments can cause performance issues. Contributing to the problem, enterprises typically move to the cloud around short, decoupled projects. They leverage whatever tools and cloud services they find attractive at the time, with no thought as to the use of common cloud services that cover commonly linked areas such as development, security, and governance.

The end result is an emerging cloud complexity crisis that could mean much of the promised value of cloud computing will not be realized. Management issues such as performance are becoming epidemic. The core questions are: How can you minimize the impact of this forthcoming complexity crisis, and what tools do you need to arm yourself with right now?

It is a difficult question for most of us to answer because we lack a standard way to deal with the performance issues that arise with cloud computing complexity, including best practices in approaches and processes, and strategic tools that work individually or as a group.

This report looks at performance issues that are unique to being in the cloud. We will explain how to understand your specific challenges, how to approach problems, and propose a reference architecture. Then we will look at available performance and management tool types, and steps to solve the issues of using those tools, including patterns of technology that can be helpful.

Key findings:

• Enterprise cloud deployments are becoming more complex. This is largely due to inadequate planning processes for cloud computing architectures, which often suffer from a focus on tactical projects versus the big picture.
• The rise in complexity presents performance issues rooted in how applications and databases communicate intra- and inter-cloud. The performance management crisis we are seeing today is the result of the increased use of multi-cloud and heterogeneous cloud services.
• There is a large array of cloud management tools with a wide variety of approaches to the complexity problem. Common patterns are beginning to emerge, along with an increased use of both automation (orchestration) and abstraction (abstractamation).
• Complexity is a problem that will get much worse before it gets better. Many enterprises will not understand the extent of the issues until they experience negative ROI from the use of cloud computing. For many enterprises, the complexity issue will take years to fix.
• The assumption is that most enterprises will not change behavior around cloud computing adoption. They will continue to opt for small tactical wins over larger strategic advantages. New approaches are needed to ultimately solve the resulting cloud performance and management problem, with a focus on fixing what is broken rather than correcting behavior from the start. In other words, vendors are waiting for enterprises to touch the stove before they supply ointment.