Manual Chapter : Investigating Server Latency Issues
Applies To:Show Versions
- 14.1.3, 14.1.2, 14.1.0
Investigating Server Latency Issues
Overview: Investigating server latency issues
This implementation describes how to investigate server latency on the BIG-IP® system. You can investigate server latency issues on the Analytics charts when Application Visibility and Reporting (AVR) is provisioned.
server latency of applications
Before you can investigate server latency, you need to have created an HTTP Analytics profile that is logging statistics internally on the BIG-IP system. The HTTP Analytics profile must be associated with one or more virtual servers, or with an iApps application service.
You can review statistics concerning server latency on the HTTP Overview chart.
Server latencyis how long it takes (in milliseconds ) from the time a request reaches the BIG-IP system, for it to proceed to the web application server, and return a response to the BIG-IP system.
- On the Main tab, click.The Overview screen opens and displays current HTTP statistics averaged over the last hour. On the screen, you can see time controls on the top, charts on the left, and a list of dimensions on the right.
- Scroll down to the Avg Server Latency chart.The chart shows the server latency for all applications and virtual servers associated with all Analytics profiles.
- To view server latency for a specific application, from the dimensions on the right, expandApplicationsand select only that application.The chart shows latency only for the selected application.
- To see more detailed latency statistics for any expanded dimensions, click the handle at the top of the dimensions column,Tables containing detailed statistics for the items in the dimensions are displayed.
- In the table, hover over the headings to see the full names of the columns, and view the data.
- Avg Server Latency (ms)shows the average server latency in milliseconds.
- Min Server Latency (ms)shows the minimum server latency in milliseconds.
- Max Server Latency (ms)shows the maximum server latency in milliseconds.
- To view the graphic charts and dimensions again, click the handle on the top left of the table.
- To view server latency for a specific virtual server, from the dimensions on the right, expandVirtual Serversand select only that virtual server.
- You can clear all filter selections or those for a dimension.
- To clear a single selection, click the selected object.
- To clear all selections, click the gear icon at the top of the column and selectClear All.
- To clear selections for a dimension, click the options icon (three horizontal lines to the left of the title), and selectClear Selection.You can click the numbered icon at the top right-hand corner of the dimension to clear all selections for a dimension.
- If further investigation is needed, select other dimensions to show latency for other entities, for example, specific pool members, URLs, countries, or client IP addresses.
If you are concerned about server latency, you can configure the HTTP Analytics profile so that it sends an alert when the average server latency exceeds a number of milliseconds for some period of time. See
Getting an alert when server latency is high.
Getting an alert when server latency is high
Before you can configure the system to send alerts concerning server latency, you need to have created an HTTP Analytics profile to collect application statistics locally (
Statistics Logging Typemust have
Internalselected). To set up email alerts, the default
analyticsprofile must specify an SMTP configuration.
You can configure the BIG-IP system to send an alert when server latency is high.
- On the Main tab, click.To have theAnalyticslisted, you need to provision Application Visibility and Reporting (AVR) first.TheProfiles: Analyticsscreen opens.
- Click the name of a previously created Analytics profile, or create a new one.
- Select theCustomcheck box.
- For theStatistics Logging Typesetting, verify thatInternalis selected. If it is not, select it.SelectingInternalcauses the system to store statistics locally, and you can view the charts on the system by starting at the Main tab, and clicking .
- To send email alerts, specify anSMTP Configuration(this can only be done on the defaultanalyticsprofile).If you created a new profile, configure SMTP later.
- For theNotification Typesetting, select how you want the system to send alerts and notifications.SyslogSelectSyslogif you want the system to send notification and alert messages to the local log system. You can view the messages on the screen.SNMPSelectSNMPif you want the system to send notification and alert messages as SNMP traps. You can create the trap by clickingConfiguration can be found here( ). Enabling SNMP automatically sets up Syslog notifications, too.SelectNotification E-Mailsfield, and clickAddto create the list. This option requires that the default analytics profile includes an SMTP configuration.When you select a notification type, the screen displays the Alerts and Notifications Configuration area, where you can indicate the criteria for alerts and notifications.
- In the Alerts and Notifications Configuration area, for theAdd New Rulesetting, define the rule that determines when the system sends an alert about server latency.
- ForAlert when, selectAverage Server Latency.
- Selectabove, and then type the number of milliseconds (the threshold) that is too high for your application. For example, type100if you want to receive an alert when latency is above 100 for 300 seconds.It is a good idea for you to get familiar with the typical average server latency of your application so you can recognize high server latency.
- SelectApplicationas the granularity level to which the threshold applies.
- ClickAdd.The rule is added to the list of Active Rules.
- If you need to configure SMTP (if sending alerts by email), click the defaultanalyticsprofile on the Profiles: Analytics screen.
- ForSMTP Configuration, select an existing configuration.
- If no SMTP configurations are listed, click theherelink to create one. When you are done, you need to select the configuration you created in the defaultanalyticsprofile.
The system sends an alert when the average server latency of an application exceeds 100 ms for 300 seconds. Another alert is sent when server latency changes back to under 100 ms for 300 seconds.