Labs 15, 16, and 17

Lab 15

For this lab I had to work with resource pools. Resource pools are used to share out resources to VMs. If all VMs are using max resources, and half are in a resource pool with less CPU shares, that half will have less performance available than in a pool with more shares. Resource pools could be used to guarantee that critical machines will have a specific amount of resources available if resources are competed.

To simulate CPU load in the virtual machines, I needed a load generator. The lab assumes you have lab files with a load generator, but I had to download one myself. I had set up internet in my network, so I just give my VMs IPs and I could access the internet to download the load generator.

VM using CPU resources:

Performance difference between resource pools. Look at the CPU usage:



If you set the amount of shares to be more similar for the resource pools, the cpu usage gets closer together for the VMs.

Lab 16

In this lab I looked at vApps. A vApp is a collection of VMs that can be managed as a single object. vApps can be used in cases where a single application requires multiple VMs such as a web server. You might want a web server, database server, and possibly some other servers that all contribute to the same service. vApps allow you to group these machines and manage their resources. You can also set the VMs to boot in a specific order with time delay between each boot to reduce load spikes on the host and to allow important services to start first.

vApps can also be stored as templates in a content library or similar as an OVF to enable the group of VMs to be deployed anywhere with ease.

vApp successfully set up:

Deleting the vApp:

Lab 17

In this lab I looked at monitoring tools. In this case the lab focused on CPU usage and CPU ready values. In the below graphs we can see the initial spike of the machines turning on, then we see a period of low use. 

The moment I started up the CPU load generators you can see the usage and CPU ready values rising. When I turn the load generators off the values drop again. 

As you can see there is a correlation between the CPU usage and CPU ready values. This is because CPU ready is the value that shows how long the VM is waiting for physical CPU resources. When the host is under load, One VM might use the CPU for one cycle, then another VM uses the CPU for another cycle. The waiting time each VM has while waiting for another VM to finish using the CPU is recorded as the CPU ready time. It can also be described as the time between when a VM is 'ready' to process, and when it actually processes.

The various monitoring tools in vSphere can be used to keep an eye on these values as well as many other values.

Monitoring. The blue circles mark where the load generators were turned off: