Considering the average VMware environment in SMB market, you will find more than 80 virtual machines in the environment, which were generated in most cases by predefined golden templates. These in turn have been configured very generous to the best of knowledge, so that the created VM will run for the next month without performance or resource problems. That means, the normal workflow in most companies looks like that the administrator gets a call from the end user, which needs one or more VMs for his daily work which will be generate from the golden templates by the administrator. A rapid growth in this case is of course inevitable, and eventually the first VMware performance problems will come very quickly.
These problems result in most cases out of a CPU overbooking. Means there are more vCPUs running on an ESX host, as a physical CPU sockets / cores are available. To put it plainly, VMs can all be operated on an ESX host, but the access time for the vCPU to the physical CPU is controlled by the ESX host. Very important to know is that a VM with 4 vCPUs needs 4 physical CPUs at the same time. Here the term "CPU Ready Time" comes into play. This value is measured in % and in most cases should not exceed 10% in the ideal case. CPU Ready Time measures the amount of time that a VM was ready to run, but had to wait for CPU. Should a user have a VM with exactly this problem, he will get back to the administrator and complain that his application on the VM or the VM itself is performing bad. Now the administrator needs to locate and resolve the problem as soon as possible, which can take a good bit of amount of time.
To support the administrator in such cases, we have developed a monitoring solution named Foglight for Virtualization Standard. The deployment takes less than 30 minutes, because this solution is based on a virtual appliance. The administrator can access the solution via the Internet Browser. After a successful configuration, this powerful solution allows you to ensure the performance and availability of your virtualization environment with actionable insights that help detect, diagnose and resolve potential issues before they impact end-user satisfaction.
For the example above regarding the vCPU overbooking, the solution provides a “Diagnose Dashboard” Fig. 1. This means that the administrator can enter the specific name of each virtual machine in the text field to check if performance problems exist.
After the VM name has been registered, it is located in the Virtual Center environment and the following resources will be analyzed:
In the diagnostic dashboard shown in Fig. 2 the detected problem is described in detail, so that the administrator can get a picture of the current situation. A chart with a definable time frame gives information when the CPU ready time value has risen. In the upper section of the dashboard the Admin find necessary workflows to solve the problem. These can be started immediately or scheduled. The workflows communicate with the VMware Virtual Center API and then forward the particular process to solve the problem. Thus, the administrator must not switch between the virtual center and the monitoring solution and saves valuable time.
A further way to analyze certain VMs offers the vScope dashboard shown in Fig. 3. With this view, the administrator receives a good overview about his entire VMware environment. All problems are highlighted, with warnings in yellow and critical issues are displayed in red.
Double-clicking on the relevant colored box automatically leads to the diagnostic dashboard, as shown in Figure 2. A longer stay on a yellow or red colored boxes will trigger the display with a summary of the identified problems.
The optimization of virtual machines
Another advantage is the optimization of virtual machines. Since many VMs are created using templates, nobody has the time to check if the allocated CPU and memory resources will also be used efficiently. The problem in this case can be, that a waste of resources based on vCPUs can quickly lead to performance problems that means high CPU ready times.
In most VMware environments, it is advisable to check first if there is a possibility to optimize virtual machines with a high number of vCPUs. Should there be machines that can be optimized by, for example, 8 vCPUs to 4 vCPUs, these resources become available back into the virtual environment and can then be used for new or existing VMs. Through this type of optimization it is possible to substantially decrease all CPU ready times.
The dashboard shown in Fig. 4, gives an overview off all virtual machines which can be optimized in vCPU resources. Foglight for Virtualization Standard
solution verifies the CPU utilization over the last 30 days as default. This can be changed by user to another suitable value. The optimization process can be performed by using the wizard shown in Fig. 5. The process can be performed scheduled or immediately. After the optimization a status mail can be send to different users.
The optimization and its result can also be applied and achieved in regards to the memory resource of any virtual machine.
Trial download of Foglight for Virtualization Standard
To learn more about his solution, the appliance can be downloaded as full version for 30 days on the following page: