If you’re thinking about moving to the cloud, you’re not alone.
From what we hear, the majority of IT departments are planning to at least take tentative steps into the cloud. They’re enticed by the promise of elastic computing for mission-critical applications, and they want the ability to only pay for what they need. They’re excited by the prospect of being able to provision for capacity solely when the application requires it, versus the bad economics of always provisioning for peak usage.
As a previous post indicated, this doesn’t mean that people will start throwing their mission-critical apps into production in the cloud. It will be a gradual deployment, starting with private clouds and moving on to hybrid clouds. But it will happen, and in greater numbers.
However, before the cloud becomes your destination, it’s necessary to first master application performance and gain complete visibility into your app. That includes having a correlated understanding of application load/throughput over time, and knowing which tiers and services need to be scaled up or down.
Doing so is the only way to get to a dynamically scaled, close-loop system–which is the end goal of cloud-based applications.
To get there, some homework is required. It’s a 3-step process for most applications to get to the point of a closed-loop system, and it’s not necessarily an easy goal to reach:
Step 1: Get full visibility into the application performance and the performance dynamics of all its moving parts, as well as how they change with load.
Step 2: Gain the ability to automate the provisioning of application services.
Step 3: Acquire the intelligence to combine the above two steps in a closed loop system.
Do steps 1 & 2 before you take your first step into the cloud–then do #3 once you’re in the cloud.
Different organizations will journey through these steps at different rates of speed and efficiency. Some will simply rearchitect their applications, or create them from scratch–but for most, it will take them some time to complete them. For some, it may even take years.
However, it’s necessary to gain visibility into the app before you gain the ability to automate provisioning. And it’s only with automated provisioning–and the intelligence to turn it into a closed-loop system–can you consider deploying the app in new environments.
Just as passing final exams is required before you get a diploma, solving application performance is a pre-requisite for moving into the cloud.