Breakdown responses time

The rate of retransmission is an indication in the form of the percentage of errors detected
during the TCP/IP exchange. Data packets that are lost or abandoned in the network are
reissued (after a timeout) by the sender (client or server) at the TCP/IP stack level. This
identification is possible because of packet numbering and sequence checking. This
phenomenon is characteristic of network congestion, increasingly longer waits,
transmission errors between some of the private or public network nodes.
Retransmission of packets impact network time and, consequently, total response time.

• Client time reflects the response time spent in the client application and to the time

required by the user (think time) to enter data and respond. Depending on the case, the
client application may be a web browser, a Java application or other. For client
applications that use the HTTP protocol, client time also includes the time spent in the
HTTP layer. Client time is dependent on how powerful the PC on which the client
application runs is, and also on the use of the applications and software running on the
machine.

• Network time corresponds to the response time between the two transport entities of thetwo computers associated with the client and the server or proxy. The "network" time
  measured represents the measurement of TCP protocol exchanges. Changes in network
  time should be analyzed against indicators of flow rates and the rate of retransmission.
• Server time represents the time spent in the part of the application that runs on servers.

This time corresponds globally to the time spent in front-end servers, application servers,
database servers, etc. For applications using the HTTP protocol, the server time includes
the time spent in the HTTP layer. (Where a proxy server is used, see the section
"Restrictions")