There have been a series of posts in the past pointing the way to various problems and solutions out in the virtualization and cloud marketplaces. Some of the topics included Hybrid Clouds, Cloud Computing itself, Apps in the Cloud, Business Alignment, Agility, and Complexity. Why have these specific topics been called out? Because a single thread (the key) to making all of this work has been missing. Analytics.
Defining Analytics:
The Wikipedia definition of Analytics is quite broad, but since our scope isn't centered on Math and Science, but instead IT (Datacenter, Cloud, Hybrid, and Virtualization, included) and that of Business (Goals, Costs, etc), we can more narrowly define it.
Analytics in these terms is the Analysis and Prediction done based on the collection of relevant data. So we have two parts to our definition, the first part is the Analysis and Prediction. Although the mathamatics is done in the background, the equally critical portion is that of the visualization of the results. Both Prediction and Analysis, whether they be in realtime or out of band, are far more valuable when they are easy to understand and can make a rapid decision on (most people like graphs and pictures as they are easier to consume). These visualizations should be available on screen, exportable, and printable as some people need to see it on paper.
Quality of Analytics:
Even the most beautiful and easy to understand (and accurate) visualizations and reports are useless without good data to fuel the calculations. That is where collecting the right data at the right time and frequency comes in. If data is collected too often, it becomes a management nightmare in and of itself! And if it isn't collected frequently enough, it isn't possible to do a thorough Analysis or make accurate Predictions (in many cases, Predictions are defined as trends going into the future).
So we've covered the WHAT and the HOW, but what about the WHY? Why are Analytics important? How can it provide the solution to Virtualization and Cloud Management?
The Key is to look at what has already happened. The need for Flexibility and the Dynamism needed to support new Applications increased demands on the Datacenter. Facing added cost pressures, the answer came back in the form of Virtualization. Virtualization led to greater Complexity, Consolidation, and Sprawl, straining resources (both people and equipment). Enter the Cloud, a way of approaching things in a service oriented manner and leading to Hybrid Cloud concepts and Cloud Bursting. This led to a need to keep track of costs and the need to be more Business Aligned.
Once businesses saw the remote possibility that IT could somehow align with business goals, the business then wanted to take advantage of it and become more Agile for a competitive advantage. Great! But how can this be maintained and the pace kept up? The simple answer is Analytics.
Just as an olympic or pro-runner trains daily and manages their calories, in order for them to win races, they can't simply bust off the starting line at full speed. Sure, they may lead in the beginning, but if they don't pace themselves, they will quickly burn out before the end of the race. They also can't eat whatever they want (take on new resources) and they won't win if they don't know their optimal pace and heart rate. The Analytic runner will study themselves, previous races, weather patterns, their shoes and diet, and they will pace themselves. They also look at trends in their running, which miles they should be running faster or slower, what zone their heart rate should be in, all in an attempt to be as efficient as possible and keep their energy at a certain level so that they can win the race.
Without a runner using Analytics, they almost never reach their full potential. In fact, most serious runners use running management tools such as GPS, Heart Rate Monitors, Weather Reports, and Training Journals. Each of these tools create lots of data, and most of it gets put online and fed into visualizations for the runner to predict how to improve their times (Time being equal to Efficiency).
Analytics for Virtualization Management and the Cloud is orders of magnitude more complex than running. And to an enterprise organization, it also proves to be much more valuable. Anyone managing virtual machines in a production environment (in the Datacenter, Cloud, or Hybrid) must have Analytics to be successful. Without proper analytics, Capacity Planning, Configuration Management, Performance Management, Chargeback or Showback may not provide you with the right information let alone the most valuable information.
Without the use of Analytics there isn't a viable way to keep up with the Chaos that is the modern Cloud and Datacenter. Even using a model based approach to management won't solve the problems of predicting and trending loads for Capacity and Performance Management. Without the right data and a way of understanding that data and interpreting the relationships within the data, Virtualization will remain impractical to manage.