12.6 DAS Statistics
This feature is for internal monitoring of your system. It is not intended for the average user. DAS Statistics monitors the following:
-
DAS_Binary
-
DAS_Core
-
Unix Communication Server
-
Collector_ Manager
-
Correlation _Engine
-
Web Server
Statistics includes the following:
-
Service: Name of the service, such as DAS_Core
-
Time: Time since the last update
-
num: Number of requests processed for this entry
-
WaitTime: Average wait time in seconds for a request before its processing starts
-
Runtime: Average time to process a request (in seconds)
-
#wait: Average size of the wait queue
-
#run: Average size of the run queue
The information is divided into three sections:
-
Requests
-
Services
-
ThreadPools
Requests keeps all the requests by channel (such as services.CorrelationService). Services does the same by service. Sometimes the information provides a breakdown by appending a category under the name, such as Services.CorrelationService or Services.RemoteObjectService.EMap.getMapPK.
For Services, the remote method calls from user-defined services (your XML services) are all under services.RemoteObjectService. Under that it puts the name of the service (such as EMap in the above example) and if asked, the name of the method (getMapPK in the above example).
When a request such as a DAS query is received by a server, a task is created and scheduled. The task is then assigned to a thread pool for execution. There can be more than one thread pool and a thread pool can service multiple services. For that reason, a request needs to wait for an available thread even if the service is not heavily used. If the statistics indicate that the wait time for a request is long and the number of requests for that service is low, check the information about the thread pools.
The numbers next to an entry are the sum for all its children. For example, requests 15 means that there are 15 requests for all requests method calls. Under that, requests.configurations 1 means that 1 of the 15 are to configurations, requests.esecurity.correlation.config 2 means that 2 of the 15 are to esecurity.correlation.config, and so on.
Figure 12-9 DAS Statistics Window
The number of requests is especially useful, because you can see where requests are going or where they are concentrated. The # waiting information is useful because it shows how busy the server is. That number should be small. If it is large, new requests (even for simple tasks) need to wait for potentially slow ones. The average run time is very important because it shows which requests are actually taking all the time, as opposed to waiting for others.