Analyze and improve data query performance
Applies To: Dynamics 365 (online), Dynamics CRM Online
You can analyze and optimize query performance using the Data Performance view, which provides an aggregated list of entities with long-running queries. A long running query is defined as a query that takes three seconds or longer to complete. Typical examples of a component that can have a long running query is a plug-in with custom FetchXML or a sub-grid or view. When you optimize an entity that has long-running queries, it could significantly reduce the amount of time the query takes to load. Behind the scenes, optimizations add one or more Microsoft SQL Server indexes.
Impact of adding or removing optimizations
Optimizations may significantly improve the performance of slow running queries. However, optimizations require storage space. Before you add an optimization carefully consider the entity to be optimized and keep the following best practices in mind.
Applying an optimization does not guarantee query performance improvement. Additionally, some optimizations can improve read performance, but cause a degradation in update performance.
Only optimize for entities that are frequently used. For example, a commonly used custom view for a sales department.
Review the impact of adding an optimization and remove optimizations that are not useful. As part of your decision to keep an optimization, you should determine if the performance tradeoff is acceptable.
Depending on the number of tables related to the entity, some entities cannot have more than twenty optimizations.
The Data Performance feature is available with Microsoft Dynamics CRM Online 2016 Update 1 and Microsoft Dynamics 365 (online).
Use the All Data Performance Logs view to apply or remove an optimization
Go to Settings > Administration > Data Performance.
In the All Data Performance Logs view, if one or more long running entity queries are detected, log items are displayed in the view. If several items appear in the view, you can sort by the Count or Optimization Impact columns to easily identify the most used queries that do not already have an optimization. Notice that out-of-box and custom entity queries that complete within three seconds aren't displayed in the view.
The columns displayed in the view provide information that can help you determine whether to add an optimization or not.
Count. Indicates the number of times a query has been executed, which indicates the popularity of the query.
Optimization Status. Indicates whether an optimization has been created for the entity. Optimization Available indicates that the entity does not have an optimization applied or that a new optimization is available for an entity that already has at least one optimization.
Optimization Impact. All potential optimizations begin with an optimization impact value of 0. Potential optimizations have no impact measurement yet because the optimization hasn't been applied. Over time, after an optimization is applied by clicking OPTIMIZE, an integer value is displayed that represents query performance impact. For example, after one week since the optimization was applied, the value that appears can help indicate whether an optimization has improved or degraded query performance for the given entity. A negative value suggests an improvement in query performance, whereas a positive number suggests a degradation in query performance. So, an optimization applied one week ago that has a value of -10% suggests a 10 percent improvement in query performance.
Select the entity that you want to create an optimization for, and then click OPTIMIZE. To remove an applied optimization, click REMOVE OPTIMIZATION.
Optimizations take effect immediately after they are applied. However, for entities that have large tables, it can take two or more hours before the optimization fully takes effect. Optimizations are processed as an asynchronous process and can be viewed in the Settings > System Jobs area of Microsoft Dynamics 365. Notice that an optimization may not be completely applied even after the system job is completed because the actual index must be built in the back-end storage engine.
© 2017 Microsoft. All rights reserved. Copyright