Azure metrics Dashboard for Azure Synapse Analytics - Part 1

This article is part 1 of a 3 part monitoring series to meet the increasing needs of customers to proactively monitor the Synapse pool resource consumption, workload patterns and other key performance metrics.

In this post, we will cover the dashboards that are available today for us on the Azure portal. These don’t require heavy customization and are very easy to set up.

To create your dashboard, navigate to the Azure Portal Home page -> Synapse Pool resource blade -> Monitoring -> Metrics

Although numerous metrics are available for building dashboards, this segment will cover the 4 most important ones for monitoring your DataWarehouse in this article. We will be using these metrics in the upcoming posts in the series as well.

1. Resource Utilization - CPU, DWU, IO percentages

From the drop down shown above choose max CPU percentage, add max DWU percentage metric and max Data IO Percentage as shown below.

Click on the pencil symbol and edit the name of the chart to your preference.

Please note that DWU Percentage and CPU percentage overlap and you may see only one of them. The DWU percentage is usually either the CPU or IO percentage, whichever is higher.

Now save the chart to a dashboard by clicking the 'pin to dashboard' option on the top righthand corner. You will be asked to choose between pinning it to an existing dashboard vs a new one as shown below.

Once the chart is saved/pinned to the dashboard, follow the same process to create the remaining charts as well, as shown below.  

2. Active and Queued queries - Concurrency details

Following the same process as above, create another chart on the same dashboard blade by adding active queries and queued queries aggregating on 'Sum'

3. Workload Group Allocation - Resource classes and their percentage allocation details

For this chart, select workload group allocation by system percent aggregating on 'Max' and split by 'Workload group'. Please note that there is a limit on the number of workload groups you can monitor.

4. Tempdb Utilization - tempdb usage across all the nodes

Add the below mentioned metric to your chart aggregating on 'Max'. It is important to note that the chart below is the minimum, average or maximum value over a 5 minute window of the average tempdb utilization across all the nodes. In general, tempdb is located on each of the nodes, however, Azure metrics do not show  the individual node level tempdb utilization as of yet. This has been brought to the attention of the development teams.

Once all the 4 charts are pinned to the dashboard, resize the charts so that they all fit on one screen like below.

Now that you have the important dashboards setup, you can build additional custom dashboards to get into more granular details about what queries/workloads are affecting your resources. This is not done by graphical user interface entirely and the second part of this post will provide you the step by step process for setting up the same.