Quick Billing Model Guide

Resource Type Billing Units

Different line items are billed in different ways. Please click on any resource type to learn more about how they are billed. If you have questions about a particular line item on your invoice, please navigate to the associated resource type section or use find (ctrl + f) to try and look up the line item from the description. 

Resource Type Descriptions

Virtual Machines

Virtual machines could be either individual VMs assigned to users or VMs that were created, managed, and destroyed by Databricks to execute jobs. Virtual machines accrue charges by the hour when they are powered on. 

Tip: Any VMs used by Databricks will use random characters as the resource name. For example, ddbdee8a1b524081a974a75d47c5f539-0 

List of sub-categories/charges

• Virtual Network Peering/Private Link – These are charges related to private, secure networking. This should be a minor cost, likely associated with Databricks VMs. 
• Virtual Machines [Size] Series – This is the hourly charge for use of the VMs, price varies based on the [Size] listed 
• Microsoft Defender CSPM – This is a charge related to security monitoring on the device 

Disk

Disk resources are the cost of data storage on a virtual machine. Disk resources are charged a flat rate for as long as the VM exists. These charges will occur regardless of whether the VM is powered on/in use. 

Tip: Similar to VMs, disks associated with Databricks VMs will end with a resource name “scratchVolume” or “containerRootVolume” (ex. ddbdee8a1b524081a974a75d47c5f539-0-scratchVolume). 

List of sub-categories / charges

• Premium or Standard / SSD or HDD – Describes the type of disk being used, this is a friendly description related to the Disk Size charges
• Disk Size [P/E/S][4/10/15] – This is a code that indicates both the type of disk [P/E/S] and the size of the disk [4/10/15/etc]. The cost of the disk will vary based on these factors.
• Disk Operations – Read/write storage charges 

Storage Account

Storage accounts are used to store data apart from Virtual Machines, similar to a file share in a traditional environment. Storage accounts are billed based on the amount of data stored, along with minor charges associated with accessing the data. 

Storage accounts can contain different types or structures of data that would be billed separately but could be associated with a single account. One example of this is your workspace shared storage account, which consists of an Azure File Share that your workspace VMs connect to, along with Data Lake / HNS storage for any jobs that need it. 

List of sub-categories / charges

• Data Stored – The cost for storing the data, this cost varies on the type of data being stored (blob, data lake, file share, etc.).
• Read/Write Operations – Costs associated with accessing or modifying data in a storage account
• Azure Data Lake Storage Gen2 Hierarchical Namespace – Storage for Databricks or other big data jobs. Enables high-performance data access.
• GRS / ZRS / LRS – This is the type of redundancy on storage account data (how and where the data is replicated across Azure datacenters). The type of redundancy will affect storage costs. 

Azure Databricks Service

Databricks is a service used to execute a large variety of analytical operations. The charges for Databricks can vary quite a bit based on the type of work being performed. It may come in the form of DBUs which are units of measurement for serverless jobs, or you may be charged for Virtual Machines and Disks if you are using compute to run your jobs. 

A Databricks Unit (DBU) is a normalized unit of processing power on the Databricks Lakehouse Platform used for measurement and pricing purposes. The number of DBUs a workload consumes is driven by processing metrics, which may include the compute resources used and the amount of data processed. 

List of sub-categories / charges

• Disk or Virtual Machine resource with a random name (ex. ddbdee8a1b524081a974a75d47c5f539-0) – These charges would show up as Virtual Machines or Disks, but are related to dynamic compute jobs in Databricks and vary based on how many VMs were created and how long they were used
• Any DBU reference – Any charges that include DBUs in the description are related to serverless jobs in Databricks; see above for the description of a DBU.
• NAT Gateway – This is a networking related charge associated with moving data in/out of Databricks for processing 

Azure Database for MySQL flexible server

These charges are related to running a MySQL database as a server-less service. This service accrues charges based on storage and how long it exists. 

List of sub-categories / charges 

• Azure Database for MySQL Flexible Server Compute – The associated compute charges for the database
• Az DB for MySQL Flexible Server Storage – Storage costs for the database
• Microsoft Defender CSPM – This is a charge related to security monitoring 

Recovery Services Vault

The Recovery Services Vault is where your data backups are stored. It is charged based on the storage of backed up data. 

Multiple copies of data are backed up at various intervals. A daily backup is kept for up to seven days, a weekly backup is stored for a month, and a monthly backup is stored for a year. This means that you could have up to 23 copies of the data (7 daily + 4 weekly + 12 monthly). 

Projects with constantly changing, large amounts of data will consume more storage and incur higher charges than projects with static data. This is related to how Azure Backup tracks/stores differences between backup intervals.