Clients always need an immediate response to their queries. Microsoft Azure fully comprehends the need of quick response time and has included the scale-out feature in Azure Analysis Services, Azure Portal, Analysis Services Rest API, and PowerShell. Scale-out feature distributes queries in different replicas and separates them in a pool which minimizes the processing time. However, it never disturbs the processing operations and performs entire work seamlessly.
During the server deployment process, only one server is deployed which works both as query server and processing server. A large number of queries disturbs the overall processing speed of the server and it usually happens when the client queries exceed the Query Processing Unit for your service plan. The availability of scale-out helps you to create a query pool for your questions and even you can generate seven replicas of the query excluding the query present in the server. It also allows you to separate the processing server from the pool and you can number your queries to meet CPU demand.
No matter how many queries are present in query replicas, the pool never disturbs the overall configuration and workloads never distribute due to the query replicas. Query replicas serve as queries against the model synchronization between the replica and the query pool. The completion of synchronization is very important for the successful completion of the process.
Metrics let to know whether or not you need a scale-out for your server. In the case you QPU maxes out daily, it is the indication that the number of queries your server is entertaining daily are more than its capability. It also happens when the query pool job length metric increases from the query thread pool.
Go to the portal and click on the scale-out option. Choose the number of queries and select ‘Separate the processing server from the query pool’ option to exclude the processing server from query server. Click Save option to save all settings.
Azure Analysis Services let to synchronize query replicas both automatically and manually. Go to Overview and choose the right model for replica from the synchronize icon.
You need to specify the server names from the overview option as there are two servers available there. Use Server Name for end client connections like a custom app, Power BI Desktop, and Excel. On the other hand, Management Server name is used for SSDT and SSMS servers.
