Service analysis timings

Service analysis operates with many different timings, describing the behavior of the service. The table below provides an overview of such timings. Timings vary in different analysis types:

DT: Distributed traces.
RT: Response time.
SF: Service flow.
MH: Method hotspots.

Regardless of which analysis you're running, it is not guaranteed that you'll see all of the timings listed here. The actual timings you see depend completely on how you're running your services. For example, if the distributed trace runs completely on one host without any networking involved, you won't see any network-related times.

Time	Appears in	Description
Response time	DT SF	The total execution time of the first node of a distributed trace. Server side The time between moments when the distributed trace is started on the server side and when the response is sent back to the client. Client side The time between moments when the client triggers the request and receives the response.
Response time	RT	The total execution time of the method. Server side The processing time of the method. Client side The time between moments when the client triggers the request and receives the response.
Processing time	DT RT	The duration of the distributed trace from start to end. It is the wall-clock time (difference between start and end time), not the sum of all asynchronous executions.
Execution time	RT	The total time taken to execute the code. It is the sum of all asynchronous executions, so it might be longer than the processing time.
Suspension	DT RT	The time during which any code execution is halted. It is usually caused by garbage collection.
Wait time	DT	The time during which the code actively waits for something (for example, `Object.wait()` or a similar functionality).
Lock wait	RT	The time during which the code is blocked. It usually caused by wait time prior to entry into a synchronized code block or by wait time to acquire a spinlock.
Active wait	RT	The time during which the code in this node is waiting for something. The wait caused by child calls is not included.
Lock time	DT	The time during which the code is blocked by waiting for a synchronous code block.
Network I/O	DT RT	The time during which the code is actively waiting for native network functions (for example, `java.net.SocketInputStream.socketRead0`). For response time analysis, the wait caused by child calls is not included.
Disk I/O	DT RT	The time during which the code is actively reading from or writing to a disk or waiting for disk input/output. For response time analysis, the wait caused by child calls is not included.
Total I/O	-	In a node, this is the sum of Network I/O and Disk I/O. When these metrics aren't available, and if the node is the direct parent of a synchronously invoked child service that isn't a custom service, the node Total I/O can be estimated as the node Duration minus the node CPU Self time.
CPU time	DT RT	The time during which the CPU executes code related to the distributed trace. The measurement is provided by OneAgent.
Self time	DT RT	The processing time of a particular node in the distributed trace.
Other	DT	Any unclassified fraction of the self time.
Elapsed time	DT	The time between executions, from when the distributed trace is created to when the method enters.
Duration	DT	Timeframe that represents the duration of a node, including its synchronous children nodes, from its start time to its end time.
Code execution	MH	Percentage of measured samples in which the method code is actively executed.
Disk I/O	MH	Percentage of measured samples in which the method is actively reading from or writing to a disk or waiting for disk input/output.
Network I/O	MH	Percentage of measured samples in which the method is actively waiting for native network functions.