============================
Monitoring bond performance
============================

Performance statistics are recorded for each bond and shown in charts on
the bond details page in the Performance section. The charts show
historical trends and patterns over a specified period and can provide
valuable insight when troubleshooting.

By default, measurements are taken every 10 seconds and reported every
60 seconds.

|image0|

Use the timeframe selection buttons to choose between displaying one
year, one month, one week, one day, six hours, one hour, or 15 minutes.
The left (<) and right (>) buttons will shift the current view, while
the forward (>>) button takes you to the latest time. When you are at
the latest time already, the forward and shift right buttons are
disabled.

You can hover your mouse over chart lines to view precise values and
timestamps. All times are displayed in your local time zone.

|image1|

As an example, the following chart shows the throughput of a bond,
broken down by the throughput of each leg. Values below the x-axis
represent download, while values above the x-axis represent upload.

|image2|

To view individual legs in more detail, you can uncheck the legs you
don't want to view.

|image3|



The following charts are shown:

#. Tunnel data traffic: the amount of inbound and outbound traffic on
   the tunnel.
#. Tunnel latency: the time taken for a ping reply to arrive over the
   tunnel. The thick line is the average, while the thin line is the
   standard deviation of the measurements.
#. Leg data traffic: the amount of inbound and outbound traffic on each
   link.
#. Leg latency: the time taken for a ping reply to arrive, as measured
   by the tunnel process. The thick line is the average for a leg, while
   the thin line is the standard deviation of the measurements.
#. Leg packet loss: the amount of packet loss on each leg.
#. Leg state changes: the rate of state changes for each leg.
#. CPU usage (average): the percent of CPU used by type (averaged across all
   cores), such as user vs system traffic.
#. CPU usage (per core): the percent of CPU used by core, such as CPU 1 vs
   CPU 2.


.. |image0| image:: /attachments/1179686/12877992.png
.. |image1| image:: /attachments/1179686/12877994.png
.. |image2| image:: /attachments/1179686/12877999.png
.. |image3| image:: /attachments/1179686/12878000.png
