Free 30 Day Trial
Find the root-cause of your cloud, hybrid-cloud
or on-prem performance issues
|
||
|
TCP Test
This test, executed by an internal agent, tracks various statistics pertaining to TCP connections to and from a host. The details of the test are provided below:
Target of the test : A host system
Agent deploying the test : An Internal agent
Outputs of the test : One set of results for each host system monitored
|
Measurement | Description | Measurement Unit | Interpretation |
---|---|---|---|
In connection rate: |
Connections per second received by the server |
Conns/Sec |
A high value can indicate an increase in input load. |
Out connection rate: |
Connections per second initiated by the server |
Conns/Sec |
A high value can indicate that one or more of the applications executing on the host have started using a number of TCP connections to some other host(s). |
Current connections: |
Currently established connections |
Number |
A sudden increase in the number of connections established on a host can indicate either an increase in load to one or more of the applications executing on the host, or that one or more of the applications are experiencing a problem (e.g., a slow down). On Microsoft Windows, the current connections metrics is the total number of TCP connections that are currently in the ESTABLISHED or CLOSE_WAIT states. The detailed diagnosis of this test, if enabled, lists the top-10 hosts that have established the maximum number of TCP connections with the monitored host. |
Connection drops: |
Rate of established TCP connections dropped from the TCP listen queue. |
Conns/Sec |
This value should be 0 for most of the time. Any non-zero value implies that one or more applications on the host are under overload or that the bandwidth of your server is insufficient. With ample bandwidth, the server can establish and serve connections before they time out. If bandwidth is insufficient, the connections fail or are dropped. |
Connection failures: |
Rate of half open TCP connections dropped from the listen queue |
Conns/Sec |
TCP counts a connection as having failed when it goes directly from sending (SYN-SENT) or receiving (SYN-RCVD) to CLOSED, or from receiving (SYN-RCVD) to listening (LISTEN). This value should be 0 for most of the time. A prolonged non-zero value can indicate either that the server is under SYN attack or that there is a problem with the network link to the server that is resulting in connections being dropped without completion. It could also indicate a bandwidth shortage. If the server has sufficient bandwidth, it can establish and serve connections before they time out. If bandwidth is insufficient, the connections fail or are dropped. This measure is not available in the Windows version of the product. |
If the test reports a high number of Current Connections, then you can use the detailed diagnosis of this measure to know which hosts are contributing the TCP connection overload on the host. The detailed diagnosis lists the IP address/host names of the top-10 hosts and the number of connections that each host has established with the monitored host.
Figure 1 : The detailed diagnosis of the Current Connections measure