We’re announcing the preview of Azure Spot Virtual Machines. Azure Spot Virtual Machines provide access to unused Azure compute capacity at deep discounts. Spot pricing is available on single Virtual Machines in addition to Virtual Machine Scale Sets (VMSS). This enables you to deploy a broader variety of workloads on Azure while enjoying access to discounted pricing. Spot Virtual Machines offer the same characteristics as a pay-as-you-go Virtual Machines, with differences in pricing and evictions. Spot Virtual Machines can be evicted anytime if Azure needs capacity.
The workloads that are ideally suited to run on Spot Virtual Machines include, but are not necessarily limited to, the following:
Preview for Spot Virtual Machines will replace the preview of Azure low-priority Virtual Machines on scale sets. Eligible low-priority Virtual Machines will be automatically transitioned over to Spot Virtual Machines. Please refer to the FAQ for additional information.
Unlike low-priority Virtual Machines, prices for Spot Virtual Machines will vary based on capacity for a size or SKU in an Azure region. Spot pricing can give you insights into the availability and demand for a given Azure Virtual Machine series and specific size in a region. The prices will change slowly to provide stabilization, thus allowing you to better manage budgets. In the Azure portal, you will have access to the current Azure Virtual Machine Spot prices to easily determine which region or Virtual Machine size best fits your needs. Spot prices are capped at pay-as-you-go prices.
Spot Virtual Machines are easy to deploy and manage. Deploying a Spot Virtual Machine is similar to configuring and deploying a regular Virtual Machine. For example, in the Azure portal, you can simply select Azure Spot Instance to deploy a Spot Virtual Machine. You can also define your maximum price for your Spot Virtual Machines. Here are two options:
There are few other options available to lower costs.
As part of this announcement, to give better flexibility, Azure is also rolling out a separate quota for Spot Virtual Machines that is separate from your pay-as-you-go Virtual Machine quota. The quota for Spot Virtual Machines and Spot VMSS instances is a single quota for all Virtual Machine sizes in a specific Azure region. This approach will give you easy access to a broader set of Virtual Machines.
Azure will try to keep your Spot Virtual Machine running and minimize evictions, but your workload should be prepared to handle evictions as runtime for an Azure Spot Virtual Machines and VMSS instances is not guaranteed. You can optionally get a 30-second eviction notice by subscribing to scheduled events. Virtual Machines can be evicted due to the following reasons:
In both scenarios, you can try to redeploy the Virtual Machine in the same region or availability zone.
Here are some effective ways to best utilize Azure Spot Virtual Machines:
The workloads that are ideally suited to run on Spot Virtual Machines include, but are not necessarily limited to, the following:
- Batch jobs.
- Workloads that can sustain and/or recover from interruptions.
- Development and test.
- Stateless applications that can use Spot Virtual Machines to scale out, opportunistically saving cost.
- Short-lived jobs which can easily be run again if the Virtual Machine is evicted.
Preview for Spot Virtual Machines will replace the preview of Azure low-priority Virtual Machines on scale sets. Eligible low-priority Virtual Machines will be automatically transitioned over to Spot Virtual Machines. Please refer to the FAQ for additional information.
Pricing
Unlike low-priority Virtual Machines, prices for Spot Virtual Machines will vary based on capacity for a size or SKU in an Azure region. Spot pricing can give you insights into the availability and demand for a given Azure Virtual Machine series and specific size in a region. The prices will change slowly to provide stabilization, thus allowing you to better manage budgets. In the Azure portal, you will have access to the current Azure Virtual Machine Spot prices to easily determine which region or Virtual Machine size best fits your needs. Spot prices are capped at pay-as-you-go prices.
Deployment
Spot Virtual Machines are easy to deploy and manage. Deploying a Spot Virtual Machine is similar to configuring and deploying a regular Virtual Machine. For example, in the Azure portal, you can simply select Azure Spot Instance to deploy a Spot Virtual Machine. You can also define your maximum price for your Spot Virtual Machines. Here are two options:
- You can choose to deploy your Spot Virtual Machines without capping the price. Azure will charge you the Spot Virtual Machine price at any given time, giving you peace of mind that your Virtual Machines will not be evicted for price reasons.
- Alternatively, you can decide to provide a specific price to stay in your budget. Azure will not charge you above the maximum price you set and will evict the Virtual Machine if the spot price rises above your defined maximum price.
There are few other options available to lower costs.
- If your workload does not require a specific Virtual Machine series and size, then you can find other Virtual Machines in the same region that may be cheaper.
- If your workload is not dependent on a specific region, then you can find a different Azure region to reduce your cost.
Quota
As part of this announcement, to give better flexibility, Azure is also rolling out a separate quota for Spot Virtual Machines that is separate from your pay-as-you-go Virtual Machine quota. The quota for Spot Virtual Machines and Spot VMSS instances is a single quota for all Virtual Machine sizes in a specific Azure region. This approach will give you easy access to a broader set of Virtual Machines.
Handling Evictions
Azure will try to keep your Spot Virtual Machine running and minimize evictions, but your workload should be prepared to handle evictions as runtime for an Azure Spot Virtual Machines and VMSS instances is not guaranteed. You can optionally get a 30-second eviction notice by subscribing to scheduled events. Virtual Machines can be evicted due to the following reasons:
- Spot prices have gone above the max price you defined for the Virtual Machine. Azure Spot Virtual Machines get evicted when the Spot price for the Virtual Machine you have chosen goes above the price you defined at the time of deployment. You can try to redeploy your Virtual Machine by changing prices.
- Azure needs to reclaim capacity.
In both scenarios, you can try to redeploy the Virtual Machine in the same region or availability zone.
Best practices
Here are some effective ways to best utilize Azure Spot Virtual Machines:
- For long-running operations, try to create checkpoints so that you can restart your workload from a previous known checkpoint to handle evictions and save time.
- In scale-out scenarios, to save costs, you can have two VMSS, where one has regular Virtual Machines and the other has Spot Virtual Machines. You can put both in the same load balancer to opportunistically scale out.
- Listen to eviction notifications in the Virtual Machine to get notified when your Virtual Machine is about to be evicted.
- If you are willing to utilize pay-as-you-go prices, then use Eviction type to «Capacity Eviction only», in the API provide "-1" as max price as Azure never charges you more than the Spot Virtual Machine price.
- To handle evictions, build a retry logic to redeploy Virtual Machines. If you do not require a specific Virtual Machine series and size, then try to deploy a different size that matches your workload needs.
- While deploying VMSS, select max spread in portal management tab or FD==1 in the API to find capacity in a zone or region.
Learn more
- Spot Virtual Machine details
- Spot Virtual Machine pricing: Windows and Linux
- Create Spot Virtual Machines in Portal
- Create Spot Virtual Machines in Azure CLI
- Create Spot Virtual Machines in Azure PowerShell
- Create Spot Virtual Machines in Azure Resource Manager templates
- Create Spot VMSS in Azure Resource Manager templates
- Planned Azure Batch support for Spot Virtual Machines
