In our last installments on the benefits of cloud computing, we started looking at the maturity of IT - people, processes and technology. Before you start making changes in the organization steam rolling toward the goal of a private cloud, it's important that you understand where you are today so that you can plot the right trajectory, and as you progress toward your goal, make course corrections as needed.
As I mentioned recently, this is not a clean-cut exercise. As we try to examine these areas in your organization and rank them, you will find areas that can be scored more mature and other areas that can be scored less mature on the same topic. This is normal. Depending on the areas of focus, improvement may vary. Let's take a look at those areas:
Last time we looked at virtualization adoption and covered areas such as:
- Availability and performance
- Business continuity
Now, let's look at a few other areas that need attention:
Management: The road to the private cloud is littered with implementations that attempt to jump to the end-goal without ensuring a good grip on their virtual infrastructure. These organizations slap a Web portal in front of their infrastructure offering "services" to their clients, but because they don't have basic management under control, resources are quickly squandered as the increase in requests for new services continues to grow. Let's look deeper at a few areas in virtual infrastructure and services management.
Resource optimization: As the frequency of requests increases, it becomes more and more important to stay on top of resources in the infrastructure. Do you regularly look for waste in the environment? Do you have a process or mechanism to remove short term or project VMs at the end of their useful life?
If you answered "yes" to those questions, you are at least ensuring that the virtual infrastructure is only serving resources to servers that need to be there. The next question is slightly tougher to handle. Are you sure that the virtual servers in your virtual infrastructure have the resources necessary to do their job without over provisioning? Too often we start our virtual servers with a particular one-size-fits-all virtual machine template and let them run until they are starved for resources. At that point, we throw more resources at it until it stops screaming. But this is a dynamic environment. Are you aware when a server no longer needs those resources? Do you have a process or mechanism to reclaim them?
Let's take that question a bit further. For servers that have never required an increase in resources, are you sure that the resources currently assigned to them are needed OR is this a case where the server is over provisioned?
A few other areas that IT needs to stay on top of are:
- Areas of infrastructure running out of capacity
- Areas of infrastructure identified for performance Improvements
- Hypervisor hosts with maxed out CPU or memory
- VMs with CPU ready bottleneck
- VMs with high CPU or memory utilization
Without a solid set of management tools in place that can monitor resources of a server and identify when it needs new resources and when it no longer needs what it has, there is no easy way to optimize the resources at hand. When an environment is small, this may not seem like a big issue but as the environment grows and as requests start coming in fast and furious, the demand for new resources will be difficult to manage.
Remember, "cloudification" isn't just about me telling my story. We can only learn and move forward by sharing ideas and exchanging concerns and questions about cloud adoption. Tell me your thoughts, experiences and best practices regarding how virtualization and the cloud have affected your enterprise and the way you deliver IT services.
Jean-Marc Seguin is chief architect for Embotics.