Report Caching in Reporting Services
A report server can cache a copy of a processed report and return that copy when a user opens the report. To a user, the only evidence available to indicate the report is a cached copy is the date and time that the report ran. If the date or time is not current and the report is not a snapshot, the report was retrieved from cache.
Caching can shorten the time required to retrieve a report if the report is large or accessed frequently. If the server is rebooted, all cached instances are reinstated when the Report Server Web service comes back online.
Caching is a performance-enhancement technique. The contents of the cache are volatile and can change as reports are added, replaced, or removed. If you require a more predictable caching strategy, you should create a report snapshot. For more information, see Setting Report Processing Properties.
Reporting Services stores temporary files in a database to support user sessions and report processing. These files are cached for internal use and to support a consistent viewing experience during a single browser session. For more information about how internal-use temporary files are cached, see Report Server Database.
A cached instance of a report is based on the intermediate format of a report. The report server generally caches one instance of a report based on the report name. However, if a report can contain different data based on query parameters, multiple versions of the report may be cached at any given time. For example, suppose you have a parameterized report that takes a region code as a parameter value. If four different users specify four unique region codes, four cached copies are created.
The first user who runs the report with a unique region code creates a cached report that contains data for that region. Subsequent users who request the report using the same region code get the cached copy.
Not all reports can be cached. If a report includes user-dependent data, prompts users for credentials, or uses Windows Authentication, it cannot be cached.
A cached report is replaced with a newer version when a user selects the report after the previously cached copy has expired. Reports that are configured to run as cached instances are removed from the cache at regular intervals based on expiration settings. You can set a report's expiration in minutes or at a scheduled time, as determined by the data's immediacy requirement. You cannot delete reports from the cache directly unless you use the SOAP API.
To configure cache expiration, you can use a shared schedule or report-specific schedule. If you use a shared schedule and it is subsequently paused, the cache does not expire while the schedule is inoperative. If the shared schedule is subsequently deleted, a copy of the schedule settings is saved as a report-specific schedule.
If a schedule expires or if the scheduling engine is unavailable at a cache expiration date, the report server runs a live report until scheduled operations can be resumed (by either extending the schedule or starting the scheduling service).
To improve server performance, you can preload the cache. To preload the cache with a collection of parameterized report instances, create a data-driven subscription that uses the Null Delivery Provider. When you specify the Null Delivery Provider as the method of delivery in the subscription, the report server targets the report server database as the delivery destination and uses a specialized rendering extension called the null rendering extension.
This feature is especially useful if you want to cache multiple instances of a parameterized report where different parameter values are used to produce different report instances. Note that you can only specify query-based parameters on the report. In contrast with other delivery extensions, the Null Delivery Provider does not have delivery settings that you can configure through a subscription definition.
When you create the data-driven subscription, you must schedule how often the reports are delivered to the cache. In order for new copies to be delivered to the cache, the old copies must have expired. Therefore, the Execution properties of the report must be configured to include cache expiration settings. The expiration setting must be consistent with the subscription schedule that you define. For example, if you create a subscription that runs every night, the cache should also expire every night prior to the subscription's run time. If the Execution properties do not include expiration times, newer deliveries are disregarded. For more information about setting properties, see Setting Report Processing Properties. For more information about using data-driven subscriptions, see Data-Driven Subscriptions.
A cached report is invalidated in response to the following events: the report definition is modified, report parameters are modified, data source credentials change, or report execution options change. If you delete a report that is stored in the cache, the cached version is also deleted.
If a report cannot be rendered from a cached instance for any reason (for example, if the parameter values that a user specifies are different from those used to produce the cached report), the report server reruns the report.