Exercise - Sizing VMs properly
Virtual machines must be sized appropriately for the expected work. A VM without the correct amount of memory or CPU will fail under load or run too slowly to be effective.
Predefined VM sizes
When you create a virtual machine, you can supply a VM size value that determines the amount of compute resources devoted to the VM, including CPU, GPU, and memory made available to the virtual machine from Azure.
Azure defines a set of predefined VM sizes for Linux and Windows from which to choose based on the expected usage.
| Type | Sizes | Description |
|---|---|---|
| General purpose | Dsv3, Dv3, DSv2, Dv2, DS, D, Av2, A0-7 | Balanced CPU-to-memory. Ideal for dev/test and small to medium applications and data solutions. |
| Compute optimized | Fs, F | High CPU-to-memory. Good for medium-traffic applications, network appliances, and batch processes. |
| Memory optimized | Esv3, Ev3, M, GS, G, DSv2, DS, Dv2, D | High memory-to-core. Great for relational databases, medium to large caches, and in-memory analytics. |
| Storage optimized | Ls | High disk throughput and IO. Ideal for big data, SQL, and NoSQL databases. |
| GPU optimized | NV, NC | Specialized VMs targeted for heavy graphic rendering and video editing. |
| High performance | H, A8-11 | Our most powerful CPU VMs with optional high-throughput network interfaces (RDMA). |
The available sizes change based on the region in which you're creating the VM. You can get a list of the available sizes using the vm list-sizes command. Try typing the following command into Azure Cloud Shell:
az vm list-sizes --location eastus --output table
Here's an abbreviated response for eastus:
MaxDataDiskCount MemoryInMb Name NumberOfCores OsDiskSizeInMb ResourceDiskSizeInMb
------------------ ------------ ---------------------- --------------- ---------------- ----------------------
2 2048 Standard_B1ms 1 1047552 4096
2 1024 Standard_B1s 1 1047552 2048
4 8192 Standard_B2ms 2 1047552 16384
4 4096 Standard_B2s 2 1047552 8192
8 16384 Standard_B4ms 4 1047552 32768
16 32768 Standard_B8ms 8 1047552 65536
4 3584 Standard_DS1_v2 1 1047552 7168
8 7168 Standard_DS2_v2 2 1047552 14336
16 14336 Standard_DS3_v2 4 1047552 28672
32 28672 Standard_DS4_v2 8 1047552 57344
64 57344 Standard_DS5_v2 16 1047552 114688
....
64 3891200 Standard_M128-32ms 128 1047552 4096000
64 3891200 Standard_M128-64ms 128 1047552 4096000
64 3891200 Standard_M128ms 128 1047552 4096000
64 2048000 Standard_M128s 128 1047552 4096000
64 1024000 Standard_M64 64 1047552 8192000
64 1792000 Standard_M64m 64 1047552 8192000
64 2048000 Standard_M128 128 1047552 16384000
64 3891200 Standard_M128m 128 1047552 16384000
Specify a size during VM creation
We didn't specify a size when we created our VM, so Azure selected a default general-purpose size for us. However, we can specify the size as part of the vm create command using the --size parameter. For example, you could use the following command to create a two-core virtual machine:
az vm create \
--resource-group "<rgn>[sandbox resource group name]</rgn>" \
--name SampleVM2 \
--image Ubuntu2204 \
--admin-username azureuser \
--generate-ssh-keys \
--verbose \
--size "Standard_DS2_v2"
Warning
Your subscription tier enforces limits on how many resources you can create, as well as the total size of those resources. Quota limits depend upon your subscription type and region. The Azure CLI lets you know when you exceed this limit with a Quota Exceeded error. If you hit this error in your own paid subscription, you can request to raise the limits associated with your paid subscription (up to 10,000 vCPUs) through a free online request.
Resize an existing VM
We can also resize an existing VM if the workload changes or if it was incorrectly sized at creation. Let's use the first VM we created, SampleVM. Before requesting a resize, we must check to see if the desired size is available in the cluster our VM is part of. We can use the vm list-vm-resize-options command:
az vm list-vm-resize-options \
--resource-group "<rgn>[sandbox resource group name]</rgn>" \
--name SampleVM \
--output table
This command returns a list of all the possible size configurations available in the resource group. If the size we want isn't available in our cluster but is available in the region, we can deallocate the VM. This command stops the running VM and removes it from the current cluster without losing any resources. We can then resize it, which re-creates the VM in a new cluster where the size configuration is available.
Note
The Microsoft Learn sandbox is limited to a few VM sizes.
To resize a VM, we'll use the vm resize command. For example, perhaps we find our VM is underpowered for the task we want it to perform. We could bump it up to a D2s_v3, where it has 2 vCores and 8 GB of memory. Type this command in Cloud Shell:
az vm resize \
--resource-group "<rgn>[sandbox resource group name]</rgn>" \
--name SampleVM \
--size Standard_D2s_v3
This command takes a few minutes to reduce the resources of the VM, and once it's done, it returns a new JSON configuration.