Monitoring SSAS
eG Enterprise provides a specialized monitoring model for SSAS.
Figure 7 : Layer model of SSAS
Each layer of Figure 7 is mapped to tests that report on the health of the operating system hosting SSAS, the network connectivity to SSAS, and a wide spectrum of SSAS-related KPIs such as thread usage, memory usage, locking, query processing, MDX scripting, and many more! Using the metrics so reported, administrators can find quick and accurate answers to the following performance queries:
-
Is SSAS available over the network? If so, is it responding quickly to network requests?
-
Is the server hosting SSAS consuming resources excessively? Is any SSAS-specific process/service responsible for this abnormal resource usage?
-
Is aggregation cache usage optimal? Is the direct cache hit ratio poor?
-
Are cache evictions occurring frequently?
-
Are connection attempts to SSAS failing frequently?
-
Are too many requests waiting for locks?
-
Was any deadlock condition observed?
-
Are too many MDX queries been run in the 'slower', 'more expensive' cell-by-cell evaluation mode?
-
Are the cell-by-cell and bulk-mode evaluation nodes using their caches effectively, or are very minimal cache hits noticed in any mode?
-
Is the SSAS instance consuming memory excessively? Should the memory thresholds be altered, so that the instance has enough memory to process its workload?
-
Is index processing slow?
-
Is sufficient memory available to process aggregations, or are temporary files used extensively to supplement the memory requirement?
-
Is the SSAS instance responding to queries quickly? If not, then what is degrading query performance - poor use of aggregations? or the ineffective usage of calculation cache, dimension cache, and/or persistent cache?
-
Are the process and query thread pools sized with adequate number of threads to run I/O and non-I/O jobs?
The MSSQL Analysis Memory Layer
