General availability: Azure Advisor recommendations for Azure Data Explorer Clusters
發佈日期： 十月 25, 2021
Azure Advisor for Azure Data Explorer is now available, providing personalized recommendations to optimize your Azure Data Explorer clusters. By analyzing your configurations and usage telemetry, Azure Advisor offers personalized, actionable recommendations that can help you reduce costs and improve performance.
Recommendation types for Azure Data Explorer:
Reduce Cost recommendations - available for clusters that can be changed to reduce cost without compromising performance. Cost recommendations include:
- Correctly size the cluster (SKU and/or instance count) to optimize cost
- Reduce cache policy for Azure Data Explorer tables
- Enable optimized auto scaling (if the cluster is really capable of scaling in)
- Delete empty and unused clusters
Stop unused clusters with data inside them
Boost Performance recommendations - help improve the performance of your Azure Data Explorer clusters. Performance recommendations include:
- Correctly size the cluster (SKU and/or instance count) to optimize performance
- Increase the cache policy or apply a different look-back period (time filter) for the queries
Operational Excellence recommendations - recommendations whose implementation does not reduce cost or imprive performance immediately but can benefit the cluster in the future.
This type surfaces recommendations for reducing the cache policy of unused tables or tables with redundant cache policies, even if the hot data saving is not significant enough to trigger an immediate cluster scale-in operation. You can think of it as a cache policy "cleaning" recommendation, meant to fine-tune the cache policy to reflect actual usage patterns.
There are two ways to find your Azure Data Explorer recommendations:
- In the Azure portal, go to your Azure Data Explorer cluster page and head directly to the "Advisor Recommendations" section in the left menu
- Access Azure Advisor resource through the Azure portal
For more information, see the full documentation.