IBM Tivoli Composite Application Manager for Transactions, Version 7.4.0.0

Enhanced network timing calculations for Web Response Time metrics

This appendix describes enhanced algorithms for calculating Web Response Time metrics, such as network, server, and client response times. The following metrics are included:

One-Way Trip Time
Total Response Time
Server Time
Network Time
Client Time
Load Time
Resolve Time

Enabling and disabling enhanced network timing calculations

You can configure these enhanced network timing calculations in the kfcmenv Web Response Time Analyzer configuration file located in one of the following directories, depending on your operating system:

On Windows systems: CANDLE_HOME\tmaitm6\wrm\analyzer\kfcmenv
On UNIX systems: CANDLE_HOME/tmaitm6/wrm/kfcmenv

To enable enhanced network timing calculations, ensure that the kfcmenv file includes the following parameter:

KFC_USE_V7201_TIMECALCS=Y

If you experience problems related to these enhanced timing calculations, you can disable this setting by setting the parameter to N:

KFC_USE_V7201_TIMECALCS=N

One-Way Trip Time

This metric represents the time for an empty packet to travel from the client to the server, or from the server to the client. The following diagram illustrates this metric:

One-Way Trip Time = (Initial ACK - Initial SYN) / 2

This is also called Zero Data Packet Time.

Total Response Time

This metric represents the time between the initial user request for the web page and the receipt of the last packet for the last page element. This can also be thought of as the time required to load the web page as experienced by the user. The following diagram illustrates this metric:

Total Response Time = (FinalReplyEnd - MainRequestStart)
 + (2 * OneWayTripTime)

Total response time starts when the client sends the start of the main request, and ends when the client receives the end of the final reply.

Accounting for overlap: The Transaction Response Time Breakdown graph in the User Sessions workspace in the Tivoli Enterprise Portal shows Client Time, Network Time, and Server Time as a stacked bar graph. However, because of overlap between Client Time and Network Time, the size of this stacked bar graph may not equal the Transaction Response Time.

Server Time

This metric represents the time taken by the server to process the request and begin to reply with data. The following diagram illustrates this metric:

Server Time = ReplyStart - RequestEnd

Server Time starts when the server has finished receiving the request for the initial HTML page, and ends when the server begins replying to the client for the initial HTML page.

Client Time

This metric is derived from the Total Response Time (see Total Response Time) and Server Time (see Server Time) and represents the difference (Client Time = Total Response Time - Server Time).

Network Time

This metric represents the time between the first response sent by the server and the last response acknowledged by the client. The following diagram illustrates this metric:

Network Time = (FinalReplyEnd - MainReplyStart) + OneWayTripTime

Network time starts when the server sends the start of the main reply, and ends when the client receives the end of the final reply.

Load Time

The following diagram illustrates this metric:

Load Time = (ReplyEnd - RequestStart)

Load Time relates to the initial HTML page and starts when the server receives the start of the current request, and ends when the server sends the end of the current reply.

Resolve Time

This metric represents the the time to load all of the page elements for the web page. The following diagram illustrates this metric:

Resolve Time = (FinalReplyEnd - SecondRequestEnd)
+ OneWayTripTime

Resolve Time starts when the server receives the end of the second request, and ends when the client receives the end of the final reply. Note that this is defined as zero if there are no secondary requests.

Last updated: September 2014