How Availability Monitoring Works

Administration › Managing Application Availability › How Availability Monitoring Works

How Availability Monitoring Works

CA Application Delivery Analysis tracks the availability of user-defined applications through passive data observation. To determine application availability, the monitoring device observes IPv4-based client activity on each server that is assigned to an application.

Availability monitoring requires:

CA Standard or Virtual Systems Monitor
CA Multi-Port Monitor

If the monitoring device observes insufficient user traffic to an application port on a server during a 5-minute interval, the monitoring device verifies availability by making active requests to the application.

Availability criteria	What it means
Less than 10 successful TCP transactions	If there are less than 10 successful TCP transactions on the application port during a 5-minute interval, the monitoring device actively checks the availability of the application.
More than 10% refused sessions	If the application port refuses 10% (or more) connection requests during a 5-minute interval, the monitoring device actively checks the availability of the application.

The monitoring device actively checks application availability by sending a TCP SYN packet to the application port on the server. For an application defined by a port range, a connection is attempted on each of the first eight ports in the range.

If the monitoring device does not receive a SYN-ACK packet response from the application server:

The application is rated Unavailable.
The monitoring device actively checks the availability of the server that hosts the application by sending a ping request to the server. Server availability is not checked unless the application is rated Unavailable. If a server:
- Responds to the ping, the server is rated as Available.
- Does not respond to the ping, the server is rated as Unavailable.

For a load-balanced application, you can choose to rate application availability based on a minimum number of assigned servers rather than checking the availability of the application on each assigned server. For example, if the load balancer distributes application traffic between a minimum of 2 and up to 5 servers, it is normal for the application to be active on at least 2 servers. To rate the application as Available, acceptable application activity must occur on at least 2 of the 5 servers. If the load-balanced application is rated Unavailable, the availability of each assigned server is checked to determine which servers are available.

More information:

Check Server Availability