Azure Databricks
Cluster coverage for auto termination
Preserve it as minimal as doable.
By default, when the cluster is created, auto termination time is 120 minutes. It must be saved as minimal as doable to scale back the price(ex. 20 minutes).
Use of spot occasion
Reference Hyperlink: https://databricks.com/weblog/2021/05/25/leverage-unused-compute-capacity-for-data-ai-with-azure-spot-instances-and-azure-databricks.html
Employee sort
Select right driver & work sort primarily based in your workload to scale back the price of the Information bricks cluster.
It is strongly recommended to make use of Normal Goal employee sort.
Information bricks job cluster as an alternative of interactive cluster in Information Manufacturing facility
Benefit of job cluster is that cluster shall be auto began and terminated when the information bricks exercise shall be triggered from the information manufacturing facility. We are able to keep away from further value of configuring job clusters as an alternative of interactive cluster as it will likely be working idle primarily based on the auto termination configuration.
Information Manufacturing facility
Take away inactive pipelines
Group creates many knowledge manufacturing facility pipelines throughout the growth. If these pipelines are scheduled then we’re paying pointless prices for these pipeline execution. Inactive pipelines must be eliminated when they don’t seem to be in use.
Disable triggers in decrease environments
A pipeline is taken into account inactive if it has no related set off or any runs throughout the month. If not required then triggers must be disabled.
Azure Digital Machines
Allow Auto shutdown function to scale back the price of VM working after enterprise hours and throughout the weekend.
Azure PostgreSQL
Efficiency configurations must be saved as minimal as doable
Normal Goal, four vCore(s), 100 GB. Altering these configuration can cut back the price of PostgreSQL drastically.
Azure SQL DB
Azure SQL DB DTU must be as little as doable to scale back the price primarily based in your workload.
App Providers
App service with Linux is extra cheaper
Price of App service configured with Linux OS is less expensive in comparison with home windows OS. When doable attempt to use App service with Linux OS.
Normal Tier is advisable
Primarily based in your workload, additionally it is advisable to configure app service with least minimal tier configuration is feasible. It is strongly recommended to make use of commonplace tier with App Service.
Azure Cosmos DB
Cosmos db request items must be as little as doable
Request items for cosmos db must be configured as little as doable. We are able to cut back a big quantity of value by protecting the cosmos db request items as little as doable.
Allow Auto Lower/Enhance RU for cosmos DB
With a purpose to optimize the price of cosmos db request items, it is suggested to find out and improve/lower cosmos db request items dynamically to extend the price financial savings whereas working with Azure Cosmos db.
Outline Price range for every useful resource & set threshold
With a purpose to proactively monitor the useful resource prices, it is suggested to find out the brink for every useful resource group and configure value finances alerts to take proactive actions to scale back the price.
