CA APM Cross-Enterprise Metrics › CA NetMaster NM for TCP/IP Metric Categories › Sockets

Sockets

CA APM Cross-Enterprise monitors data for the following Sockets metrics from a connected CA NetMaster NM for TCP/IP region:

Active Conns

Displays the current number of active connections to this TCP server port, summed for all monitored stacks.

Input Bytes/Sec

Displays the rate of input to this TCP server port in bytes per second over the last full minute.

Output Bytes/Sec

Displays the rate of output from this TCP server port in bytes per second over the last full minute.

Backlog Q Depth

Displays the current number of requests in the TCP backlog queue.

Avg Appl Response (10 ms)

Displays the time between the following events:

• When the local application replied with the first ACK to a received request
• When the local application then sends the next data packet to respond to the request

An average of its local application response time is maintained for each TCP connection. Then the average of these values is taken for all concurrently active TCP connections with this port that have had packet activity within the last few minutes. This value is recalculated every 5 minutes.

Note: This metric is most meaningful with TCP applications that communicate in a regular request-response pattern, such as HTTP.

The measurement unit is 10 ms (1/100th second), that is, 234 = 2.34 seconds.

Time to 1st Response (10 ms)

Displays the time between the following events:

• When the local application replied with the first ACK to a received request
• When the local application then sends the next data packet to respond to the request

This metric is equivalent to the time to first response metrics other CA APM products provide.

The local average application response time is continually measured and averaged for every turn in the TCP connection. In contrast, the time to first response is only measured once per TCP connection. The average of these values is then taken for all concurrently active TCP connections with this port that have had packet activity within the last few minutes. This value is recalculated every 5 minutes.

Note: This metric is most meaningful with TCP applications that communicate in a regular request-response pattern, such as HTTP.

The measurement unit is 10 ms (1/100th second), that is, 234 = 2.34 seconds.

For detailed diagnosis of TCP server port activities, use these functions in the CA NetMaster NM for TCP/IP region:

• Use ASMON monitoring to alert on port attributes.
• Set up business applications to split connections to this port by remote addresses or to combine this port traffic with related ports.
• Set up packet-based events for real-time notification of critical TCP port connection, workload, fragmentation, and error activities.
• Use SmartTrace to for real-time packet stream viewing and deep packet inspection of the traffic flowing over a specific connection with a port.
• Use multiple TCP tracing in SmartTrace to trace different connections with a port separately.