About Web Response Time
Web Response Time uses network monitoring to capture HTTP/HTTPS transaction data such as response time and status codes. You can use it to capture performance and availability data of actual users for Service Level Agreement (SLA) reporting.
- Monitor user performance and availability for web-based applications.
- Capture web request response time and its segmentation
- Monitor the performance of web page request and each embedded object in that web page. This feature, which can be switched on or off, can identify if any graphics, tables, JavaScript, or Applets are causing response time problems. Audio and video request monitoring is not available.
- Monitor the response time down to and including the workstation without being physically on the workstation
- Provide remote or local monitoring of HTTP and HTTPS for any type of server, using Appliance Mode. If Web Response Time cannot run on a particular computer (for example, because of security concerns or because the operating system is not supported by Web Response Time), it can run on a separate computer, called the appliance computer.
- Monitor a specific network interface. By default, the Web Response Time Analyzer monitors all network interfaces, however, you can limit the monitoring to one network interface.
Web Response Time parses and then stores data and presents it to the Tivoli Enterprise Monitoring Server as shown in the following illustration:
- Web Response Time Analyzer
- The Web Response Time Analyzer detects
and tracks network transactions between the web server and users or
other computers, and calculates response time for each transaction.
The KT5Agent is a Web Response Time Analyzer application. It defines monitoring criteria to the analyzer through the KFC1 API. The analyzer creates a filter corresponding to the monitoring requests. The application can define as many filters as needed, and different application filters might be either identical or similar.
The Web Response Time Analyzer processes captured data per network and application protocols and creates transaction response time data records. Every transaction response time data record is examined by every filter for its applicability. When a transaction response time data record passes a filter, the response time data attributes are formatted into a transmission buffer and delivered to the owning application. Running response time data records through all filter criteria and formatting all transaction attributes and timestamps into transmission buffers is a time-consuming process.
The Web Response Time Analyzer monitors individual web page response times and delivers the data to the KT5Agent.
- KT5Agent
- The KT5Agent aggregates the response time data provided by the Web Response Time Analyzer into application, client, and server-centric summary data.
Web Response Time uses network monitoring to capture HTTP/HTTPS transaction data such as response time and status codes, as shown in the following illustration.
The metrics help determine the quality of service of web-based applications, as experienced by customers and other requestors. Those requestors can be users visiting web sites or computers performing application-to-application or business-to-business processes.
- How it works
- The Web Response Time agent
listens on the network port to collect data transportation performance
and then calculates web application layer performance by postprocessing.
It tracks application performance from a transaction-oriented point
of view and collects performance information by organizing business
critical processing into transactions.
The agent is on the monitoring server computer, not on the users' workstations. It does not interfere with network data flow, affect network performance, or alter transmitted data; and it does not identify individual users. The agent examines network transmission data and measures transaction response time based on network and application protocols.
- Network protocols
- The agent calculates response time for web transactions that use the TCP/IP protocol. Network protocol response time is the elapsed time of the data exchange between a network addressable source and a network addressable destination. The agent interprets specific protocol details and determines whether each request and its subsequent data replies were delivered and acknowledged. By monitoring this data exchange cycle, the agent calculates network protocol response time and collects other important information.
- Application protocols
- Network protocol response time is only part of the picture. Each web application transaction; such as downloading a file, retrieving a web page, or performing a database inquiry; represents a unit of work, with tangible business value to customers. Therefore, the agent further processes and consolidates response time measurements, based on specific application protocols. The resulting application response time considers the possibility of parallel and concurrent network transaction activities. For example, because the page and its embedded objects are retrieved in parallel, the response time for a web page is the total of the response time for the page itself and the response times of all the objects (such as graphics) embedded in the page. The agent arrives at an accurate figure for total response time for the page, and its figures for the response time of each embedded object tell you which objects are causing page download delays.
- HTTP Transactions
- The agent can monitor HTTP transactions without any modification
of the web server configuration. The data log contains separate records
for each web page and for each object embedded in that web page. The agent generates a web object record only if the object is served by the monitored web server. Even if a web page comes from the monitored web server:
- Some embedded objects might come from other servers. In that case, the Collector creates log records for the web page and for the objects from the monitored server, but not for the objects from the other servers.
- The Collector might not create an object record for it if the object takes a long time to retrieve. In that case, the object is excluded from the total response time calculation for the page. Later, when the browser retrieves the object, the Collector generates a web page record for it, rather than an object record.
- How Web Response Time calculates response time
- Unless configured otherwise, the agent creates only a response
time for steps 1 and 2. If Web Response Time is
configured to monitor objects, it examines the object Multipurpose
Internet Mail Extensions (MIME) type delivered by the server. If the
MIME type specification is not present in the data stream, Web Response Time uses
the file name extension (for example, .gif, .jpg,
or .js) as a guide for setting the object type.
For more information, see Enhanced network timing calculations for Web Response Time metrics.