Statistics about response

Item	Details
RPC type	Number of RPCs issued to the applicable service from the node that outputs the journal, for each type
Response time^#1	Length of time between the point when dc_rpc_call is issued and the point when the response text is received. However, the length of time differs depending on the flag as shown below: DCNOFLAGS (synchronous-response RPC) Length of time between the point when dc_rpc_call is issued and the point when the response text is received DCRPC_CHAINED (chained RPC) Length of time between the point when dc_rpc_call is issued and the point when the response text is received DCRPC_NOWAIT (asynchronous-response RPC) Length of time between the point when dc_rpc_call is issued and the point when the response is written in the out area DCRPC_NOREPLY (non-response RPC) No response is acquired.
Service execution time^#2	Length of time required to execute a service function. This information is not acquired for non-response RPCs.
Service wait time^#2	Length of time between the point at which a request is collected in the schedule queue and the point at which the request is fetched from the schedule queue. This information is not acquired for non-response RPCs.

Response statistics are output differently depending on the execution type of each service. The following shows an example output of statistics for each type of execution.

(a) When the target service exists in a node different from the node that requests the service

The following figure shows a case when the target service exists in a node different from the node that requests the service.

Figure E-1 When the target service exists in a node different from the node that requests the service

[Figure]

When you edit the journals acquired in the hosts in Figure D-1, the response statistics are output as follows for each host:

Edited journal of host A

#1: Values of t10 to t1 in Figure D-1

#2: Values of t9 to t3 in Figure D-1

#3: Values of t3 and t2 in Figure D-1

The details about each output item are as follows:

Number of events for each RPC type: Host A issued a synchronous-response RPC to service A in service group A once.

Response time: Length of time between the point when the SUP on host A issues dc_rpc_call and the point when the SUP receives a response

Execution of user service (service execution time): Length of time between the point when service A in service group A starts processing and the point when service A sends a response. However, the communication time for sending the response is not included.

Service wait (service wait time): Length of time between the point when the SUP on host A enters a request in the schedule queue and the point when the request is fetched from the schedule queue

Edited journal of host B

#1: 0

#2: Values of t9 to t3 in Figure D-1

#3: Values of t3 and t2 in Figure D-1

#4: Values of t8 to t4 in Figure D-1

#5: Values of t7 and t6 in Figure D-1

#6: Values of t6 and t5 in Figure D-1

The details about each output item are as follows:

Service A in service group A

Number of events for each RPC type: Host B has never called service A in service group A.

Response time: Length of time between the point when the SUP on host A issues dc_rpc_call and the point when the SUP receives a response. This value cannot be acquired since the journal of host B is edited. Therefore, 0 is written.

Execution of user service (service execution time): Length of time between the point when service A in service group A starts processing and the point when service A sends a response. However, the communication time for sending the response is not included.

Service wait (service wait time): Length of time between the point when the SUP on host A enters a request in the schedule queue and the point when the request is fetched from the schedule queue

Service B in service group B

Number of events for each RPC type: Host B issued a synchronous-response RPC to service B in service group B once.

Response time: Length of time between the point when SPP A on host B issues dc_rpc_call and the point when SPP A receives a response

Execution of user service (service execution time): Length of time between the point when service B in service group B starts processing and the point when service B sends a response. However, the communication time for sending the response is not included.

Service wait (service wait time): Length of time between the point when SPP A on host B enters a request in the schedule queue and the point when the request is fetched from the schedule queue

Edited journal of host C

#1: 0

#2: Values of t9 to t3 in Figure D-1

#3: Values of t3 and t2 in Figure D-1

The details about each output item are as follows:

Number of events for each RPC type: Host C has never called service B in service group B.

Response time: Length of time between the point when SPP A on host B issues dc_rpc_call and the point when SPP A receives a response. However, this value cannot be acquired since the journal of host C is edited. Therefore, 0 is written.

Execution of user service (service execution time): Length of time between the point when service B in service group B starts processing and the point when service B sends a response. However, the communication time for sending the response is not included.

Service wait (service wait time): Length of time between the point when SPP A on host B enters a request in the schedule queue and the point when the request is fetched from the schedule queue

The relationship between the journals of the hosts and the statistics acquired from the journals is as follows:

When you edit the journal of host A, you can acquire the statistics on SPP A operating in host B.
When you edit the journal of host B, you can acquire the statistics on SPP B operating in host C.
When you edit the journal of host B, you cannot acquire the statistics on SPP A.
When you edit the journal of host C, you cannot acquire the statistics on SPP B.

(b) When a service is executed in the local node

The following figure shows a case of executing a service in the local node.

Figure E-2 When a service is executed in the local node

[Figure]

When you edit the journal that is acquired in the host in Figure D-2, the response statistics are output as follows:

[Figure]

#1: Values of t10 to t1 in Figure D-2

#2: Values of t9 to t3 in Figure D-2

#3: Values of t3 and t2 in Figure D-2

#4: Values of t8 to t4 in Figure D-2

#5: Values of t7 and t6 in Figure D-2

#6: Values of t6 and t5 in Figure D-2

The details about each output item are as follows:

Service A in service group A

Number of events for each RPC type: The local node issued a synchronous-response RPC to service A in service group A once.

Response time: Length of time between the point when the SUP issues dc_rpc_call and the point when the SUP receives a response

Execution of user service (service execution time): Length of time between the point when service A in service group A starts processing and the point when service A sends a response. However, the communication time for sending the response is not included.; The number of events is 2 since rpc_response_statistics=Y is specified for both the SUP and SPP A.; The average value is determined by dividing (maximum value + minimum value) by 2.

Service wait (service wait time): Length of time between the point when the SUP enters a request in the schedule queue and the point when the request is fetched from the schedule queue; The number of events is 2 since rpc_response_statistics=Y is specified for both the SUP and SPP A.; The average value is determined by dividing (maximum value + minimum value) by 2.

Service B in service group B

Number of events for each RPC event: The local node issued a synchronous-response RPC to service B in service group B once.

Response time: Length of time between the point when SPP A issues dc_rpc_call and the point when SPP A receives a response

Execution of user service (service execution time): Length of time between the point when service B in service group B starts processing and the point when service B sends a response. However, the communication time for sending the response is not included.; The number of events is 2 since rpc_response_statistics=Y is specified for both SPP A and SPP B.; The average value is determined by dividing (maximum value + minimum value) by 2.

Service wait (service wait time): Length of time between the point when SPP A enters a request in the schedule queue and the point when the request is fetched from the schedule queue; The number of events is 2 since rpc_response_statistics=Y is specified for both SPP A and SPP B.; The average value is determined by dividing (maximum value + minimum value) by 2.

(c) When a service is executed from a client

The following figure shows a case when a service is executed from a client.

Figure E-3 When a service is executed from a client

[Figure]

When you edit the journal acquired in the host in Figure D-3, the response statistics are output as follows:

[Figure]

#1: 0

#2: Values of t9 to t3 in Figure D-3

#3: Values of t3 and t2 in Figure D-3

#4: Values of t8 to t4 in Figure D-3

#5: Values of t7 and t6 in Figure D-3

#6: Values of t6 and t5 in Figure D-3

The details about each output item are as follows:

Service A in service group A

Number of events for each RPC type: The local node has never called service A in service group A. Even if the client is in the same node, calls from the client cannot be included in the count since clients cannot acquire response statistics.

Response time: Length of time between the point when the CUP issues dc_rpc_call and the point when the CUP receives a response. However, this value cannot be acquired since clients cannot acquire response statistics. Therefore, 0 is written.

Execution of user service (service execution time): Length of time between the point when service A in service group A starts processing and the point when service A sends a response. However, the communication time for sending the response is not included.

Service wait (service wait time): Length of time between the point when the CUP enters a request in the schedule queue and the point when the request is fetched from the schedule queue

Service B in service group B

Number of events for each RPC type: The local node issued a synchronous-response RPC to service B in service group B once.

Response time: Length of time between the point when SPP A issues dc_rpc_call and the point when SPP A receives a response

Execution of user service (service execution time): Length of time between the point when service B in service group B starts processing and the point when service B sends a response. However, the communication time for sending the response is not included.; The number of events is 2 since rpc_response_statistics=Y is specified for both SPP A and SPP B.; The average value is determined by dividing (maximum value + minimum value) by 2.

Service wait (service wait time): Length of time between the point when SPP A enters a request in the schedule queue and the point when the request is fetched from the schedule queue; The number of events is 2 since rpc_response_statistics=Y is specified for both SPP A and SPP B.; The average value is determined by dividing (maximum value + minimum value) by 2.

Appendix E.3 Statistics about response

(1) Details about response statistics

(2) Output example of response statistics

(a) When the target service exists in a node different from the node that requests the service

(b) When a service is executed in the local node

(c) When a service is executed from a client