When monitoring for potential network bottlenecks, items such as the network bandwidth that is used on a network segment and the rate at which the server is transmitting bytes through the network can be monitored. Use the following objects and counters to identify network bottlenecks.
Network Segment ObjectRedirector Object
For the Network Segment Object, monitor the counter described below:
% Net Utilization
Shows the percentage of network bandwidth in use on a network segment. The threshold must be determined based on the type of network used. For example, for Ethernet networks, 30 percent is the recommended threshold.
For the Redirector object, monitor the counters described below:
Bytes Total / secNetwork Errors / sec
Provides the number of bytes per second sent and received by the network redirector. By comparing the maximum throughput of the network card with the maximum value of this counter, this counter determines if the network traffic is a bottleneck in the system.
Measures the number of unexpected errors the redirector receives. If network problems are suspected, verify that this counter is above zero. If it is above zero, the system log may be helpful by providing details on the network error.Server Object
For the Server Object, monitor the counters described below:
Work Item Shortages
Bytes Total / sec
Shows the rate at which the server is transmitting bytes through the network. If the sum of Bytes Total / sec for all servers is almost equal to the maximum transfer rates of the network, then this network may need to be segmented.
Shows the number of times STATUS_DATA_NOT_ACCEPTED is returned. This occurs when no work item is available or can be allocated to service the incoming request. If the value reaches approximately 3, consideration should be taken to tuning InitWorkItems or MaxWorkItems entries in the Registry (in HKEY_LOCAL_MACHINE\SYSTEM\
Pool Paged Peak
Indicates the proper sizes for the paging file and the amount of physical memory.Server Work Queues Object
For the Server Work Queue object, monitor the counters described below:
Shows the current length of the server work queue for a processor. If the value reaches approximately 4, there may be a processor bottleneck. Queue Length is an instantaneous counter. This value needs to be observed over several intervals to determine whether there is a problem.