This article is contributed. See the original author and article here.
We are excited to announce that Azure Spot VMs are now supported in Azure Databricks. Together, Spot VMs and Azure Databricks help innovative customers like aluminium and energy producer Hydro to accelerate data + AI workloads while optimizing costs. This will allow you to access unused Azure compute capacity at deep discounts—up to 90 percent compared to pay-as-you-go prices.* This feature is GA in all regions where Azure Databricks is available today.
How can you use and save cost using this feature? With Azure Spot VMs, you can access unused Azure compute capacity at deep discounts. Discounts vary based on region, VM type, and availability of compute capacity when your workload is deployed. You pay up to the maximum price that you optionally agree to in advance. You can specify the use of Spot VMs for your clusters and pools via the Databricks UI or Rest API.
Create an Azure Databricks cluster with Spot VMs using the UI
When you create an Azure Databricks cluster, select your desired instance type, Databricks Runtime version and then select the “Spot Instances” checkbox as highlighted below.
Spot VMs are ideal for workloads that can be interrupted, providing scalability while reducing costs. Azure Databricks automatically handles the termination of Spot VMs by starting new pay-as-you-go worker nodes to guarantee your jobs will eventually complete. This provides predictability, while helping to lower costs.
When a cluster is created with Spot instances, Databricks will allocate Spot VMs for all worker nodes, if available. The driver node is always an On-Demand VM. During your workload runs, Spot VMs can be evicted when Azure no longer has available compute capacity and must reallocate its resources. Spot VMs will also be evicted when the current price exceeds the maximum price that you optionally agreed to pay before the VMs were allocated. In either case, pay-as-you-go instances will automatically replace any evicted Spot VMs.
Create an Azure Databricks cluster with Spot VMs using the REST API
With the Azure Databricks Clusters REST API, you have the ability to choose your maximum Spot price and fallback option if Spot instances are not available or are above your maximum price. Create a bearer token in the Databricks UI. This will be used to authenticate when making your API call.
Create an Azure Databricks warm pool with Spot VMs using the UI
You can use Azure Spot VMs to configure warm pools. Clusters in the pool will launch with spot instances for all nodes, driver and worker nodes. When creating a pool, select the desired instance size and Databricks Runtime version, then choose “All Spot” from the On-demand/Spot option.
At any point in time when Azure needs the capacity back, the Azure infrastructure will evict Azure Spot Virtual Machines with 30 seconds notice. Databricks will attempt to replace the evicted Spot Virtual Machines, if unsuccessful, on-demand VMs are deployed to replace evicted Spot VMs.
Create a warm pool with Spot VMs using the Instance Pools API
The Instance Pools API can be used to create warm Azure Databricks pools with Spot VMs. In addition to the options available in the Azure Databricks UI, the Instance Pools API enables you to specify a maximum Spot VM price and fallback behavior if Spot VM capacity is unavailable.
Spot Pricing, Trends and Eviction Rates
Spot VM pricing can be viewed for all series on the Linux VM pricing page. The last column in the tables under each series type shows the savings based on region.
On the Databricks pricing page, you can view the ‘Pay As You Go Total Price’ (PAYG) and ‘Spot (%Savings) Total Price’. Note that the DBU price does not change with the use of Spot VMs. Savings are from discounts on Spot VMs only. Please note that Spot VMs pricing does not include network, storage or other resources, which are billed separately.
- Pay As You Go Total Price = Pay As You Go Price + DBU Price
- Spot Total Price = Spot VM Price (% Savings of PAYG) + DBU Price
Customers can also look at Spot VM pricing trends for the last 3 months and eviction rate. To see this, navigate to the Create a virtual machine page within the Azure Portal and click the “View pricing history and compare pricing in nearby regions” link. This will show you historical pricing and eviction rate for the regions and instances you select.
Learn more about using Azure Spot VMs with Azure Databricks by viewing the documentation: Azure Spot VMs, Azure Databricks Clusters, Azure Databricks Clusters API, Azure Databricks pools and Instance Pools API . To get started with Azure Databricks, visit the Azure Databricks page and attend up-coming Azure Databricks events.
*Actual discounts may vary based on region, VM type, and Azure compute capacity available when the workload is deployed. Please note that Spot VMs pricing does not include network, storage or other resources, which are billed separately.
Brought to you by Dr. Ware, Microsoft Office 365 Silver Partner, Charleston SC.