doAzureParallel/docs/31-vm-sizes.md

Virtual Machine Sizes

The doAzureParallel package lets you choose the VMs that your code runs on giving you full control over your infrastructure. By default, we start you on an economical, general-purpose VM size called "Standard_A1_v2".

Each doAzureParallel pool can only comprise of of a collection of one VM size that is selected upon pool creation. Once the pool is created, users cannot change the VM size unless they plan on reprovisioning another pool.

Setting your VM size

The VM size is set in the configuration JSON file that is passed into the registerPool() method. To set your desired VM size, simply edit the vmSize key in the JSON:

{
  ...
  "vmSize": <Your Desired VM Size>,
  ...
}

Choosing your VM Size

Azure has a wide variety of VMs that you can choose from.

VM Categories

The three recommended VM categories for the doAzureParallel package are:

• Av2-Series VMs
• F-Series VMs
• Dv2-Series VMs

Each VM category also has a variety of VM sizes (see table below).

Generally speaking, the F-Series VM is ideal for compute intensive workloads, the Dv2-Series VMs are ideal for memory intensive workloads, and finally the Av2-Series VMs are economical, general-purpose VMs.

The Dv2-Series VMs and F-Series VMs use the 2.4 GHz Intel Xeon® E5-2673 v3 (Haswell) processor.

VM Size Table

Please see the below table for a curated list of VM types:

VM Category	VM Size	Cores	Memory (GB)
Av2-Series	Standard_A4_v2	4	8
Av2-Series	Standard_A8_v2	8	16
Av2-Series	Standard_A2m_v2	2	16
Av2-Series	Standard_A4m_v2	4	32
Av2-Series	Standard_A8m_v2	8	64
F-Series	Standard_F1	1	2
F-Series	Standard_F2	2	4
F-Series	Standard_F4	4	8
F-Series	Standard_F8	8	16
F-Series	Standard_F16	16	32
Dv2-Series	Standard_D1_v2	1	3.5
Dv2-Series	Standard_D2_v2	2	7
Dv2-Series	Standard_D3_v2	4	14
Dv2-Series	Standard_D4_v2	8	28
Dv2-Series	Standard_D5_v2	16	56
Dv2-Series	Standard_D11_v2	2	14
Dv2-Series	Standard_D12_v2	4	28
Dv2-Series	Standard_D13_v2	8	56
Dv2-Series	Standard_D14_v2	16	112

The list above covers most scenarios that run R jobs. For special scenarios (such as GPU accelerated R code) please see the full list of available VM sizes by visiting the Azure VM Linux Sizes page here.

To get a sense of what each VM costs, please visit the Azure Virtual Machine pricing page here.

Low Priority VMs

Low-priority VMs are a way to obtain and consume Azure compute at a much lower price using Azure Batch. Since doAzureParallel is built on top of Azure Batch, this package is able to take advantage of low-priority VMs and allocate compute resources from Azure's surplus capacity at up to 80% discount.

Low-priority VMs come with the understanding that when you request it, there is the possibility that we'll need to take some or all of it back. Hence the name low-priority - VMs may not be allocated or may be preempted due to higher priority allocations, which equate to full-priced VMs that have an SLA.

And as the name suggests, this significant cost reduction is ideal for low priority workloads that do not have a strict performance requirement.

With Azure Batch's first-class support for low-priority VMs, you can use them in conjunction with normal on-demand VMs (dedicated VMs) and enable job cost to be balanced with job execution flexibility:

• Batch pools can contain both on-demand nodes and low-priority nodes. The two types can be independently scaled, either explicitly with the resize operation or automatically using auto-scale. Different configurations can be used, such as maximizing cost savings by always using low-priority nodes or spinning up on-demand nodes at full price, to maintain capacity by replacing any preempted low-priority nodes.
• If any low-priority nodes are preempted, then Batch will automatically attempt to replace the lost capacity, continually seeking to maintain the target amount of low-priority capacity in the pool.
• If tasks are interrupted when the node on which it is running is preempted, then the tasks are automatically re-queued to be re-run.

For more information about low-priority VMs, please visit the documentation.

You can also check out information on low-priority pricing here.