eG Monitoring
 

Measures reported by LSDBStoreTest

Microsoft Lync server performance is affected by various factors such as user profiles, system architecture, software, hardware components, third-party integration points such as gateways and telephony equipment, network connectivity and performance, Windows Active Directory service configuration and performance in addition to the Windows operating system functionality. At the core of the Lync server deployments' performance is the server software and hardware it is implemented on. As an example, a front-end server must have sufficient hardware resources to cope with the expected (short-term) user load. If a respective front-end server is required to provide services to 10 thousand users, then an adequately configured server must meet the expected load requirements to ultimately help ensure the best possible end-user experience.

Monitoring server performance is therefore extremely important to gauge whether the implemented server infrastructure have suitable hardware resources for the day-to-day peak-load requirements. Monitoring server performance helps identify system bottlenecks allowing administrators to apply corrective action before the end-user experience is affected. The performance data obtained is then used for long-term capacity planning. For a server performance to ne extremely high, it is mandatory to track the performance of the backend SQL server database server used by the front end pool of the Microsoft Lync server deployment. This is exactly where the LSDBStoreTest test helps!

This test monitors the backend database server and proactiviely alerts the administrators if the following exceeds the optimal range:

  • The time spent by the requests in the queue while waiting to contact the backend server
  • The time taken by the backend server to process a request and
  • The rate at which the requests were throttled

The measures made by this test are as follows:

Measurement Description Measurement Unit Interpretation
QueueDepth Indicates the number of requests that are waiting in the queue to be run. Number The value of this measure should be less than 500. A value above 500 indicates that the backend database is busy and is unale to respond to requests quickly.
QueueLatency Indicates the time spent by a request in the queue while waiting to contact the backend server. Secs A low value is desired for this measure. A gradual/sudden increase in the value of this measure indicates that the resource utilization of the backend server is high. Also, there may be a high network latency between the frontend and backend servers.
SprocLatency Indicates the time taken by the backend server to process a request. Secs A low value is desired for this measure. If both the SprocLatency and QueueLatency measures are high, then it indicates a bottleneck in the SQL Server back-end database. The disk IO contention and the SQL Server database active transactions are good indicators to identify the database that is having more contention than other databases.
ThrottledReq Indicates the rate at which the requests were rejected after being retried. Request/Sec The value of this measure will be high if the value of the QueueLatency measure is abnormally high.