RTM Overview

Packeteer's response time measurement (RTM) enables you to measure the quality of users' network experience. RTM provides performance statistics, threshold monitoring, high-level problem indicators, and performance graphs. You can use the response-time analysis to highlight problems and confirm the effectiveness of solutions.

Note: The response time measurement feature is not available on PacketShaper ISP models or the PacketShaper 1200 model.

RTM Functions

The RTM feature performs the following functions:

• Tracks delay statistics for connection-based TCP traffic classes (for example, you can measure response times for applications, web traffic, individual hosts, or subnets). See Monitor Response Times.
• Breaks each response time measurement into network delay, server delay, and total delay. See Show Transaction Delays.
• Identifies the clients and servers with the slowest performance records. See Enable Worst Client and Server Analysis.
• Allows you to set acceptability standards and track whether performance meets those standards. See Set Thresholds.
• Offers current and historical performance data in tables and graphs, as well as in a MIB (management information base)

The PacketShaper is positioned in the corporate network to monitor and control all the traffic that passes through it. This gives PacketWise an opportunity to provide accurate response time measurements at low cost. Since PacketWise already handles and classifies every packet, it can calculate the time traffic spends traveling between a client and a server. Instead of polling for response data, PacketWise notes response times as traffic passes through the unit.

Benefits

Response time measurement quantifies what has traditionally been subjective information. Performance measurement allows you to:

• Recognize performance problems before they impact business
• Assess the results of configuration changes, including changes to PacketWise's policy and partition configuration
• Provide and monitor meaningful Service Level Agreements (SLAs)
• Feed strategies for bandwidth management, server balancing, and topology planning

How Response Times are Tracked

Traffic class response times are tracked if the associated traffic is TCP and conforms to recognized transaction behavior, including request/response pairs, push bits, MSS (maximum segment size), and so on. PacketWise stores at least one day’s worth of RTM statistics at one-minute intervals and at least one month’s worth of data at one-hour intervals.

Response times are recorded only in the client-to-server direction. If there is no traffic in this direction, the RTM values recorded will be zero. Note that the Monitor Response Time screen in the browser interface and the rtm show command in the CLI list only classes that have RTM data.

Application response times involve bi-directional traffic. After all, it hardly makes sense to track a request without tracking its corresponding response. When looking at RTM statistics, you need to determine which statistics represent the activity you are trying to capture.

For example, suppose you want performance data for your branch office accessing web servers that are located either at corporate headquarters or other companies. This traffic is represented by the /Inbound/HTTP class and the /Outbound/HTTP class. In this case, the response times that tell the performance story are displayed in the /Outbound/HTTP traffic class because it represents client-to-server traffic. While the relevant RTM data is in the /Outbound/HTTP class, you would most likely apply a policy to the /Inbound/HTTP class, because it represents the download of data to the client.

See also:

Analyze Application Response Times

Troubleshooting Response Time Measurement Issues

Response Time Measurement's Technical Details