The BIG-IP® system includes a feature known as fast failover. Fast failover is a feature that is based on the concept of an HA group. An HA group is a set of trunks, pools, or clusters (or any combination of these) that you want the BIG-IP system to use to calculate an overall health score for a device in a redundant system configuration. A health score is based on the number of members that are currently available for any trunks, pools, and clusters in the HA group, combined with a weight that you assign to each trunk, pool, and cluster. The device that has the best overall score at any given time becomes or remains the active device.
The fast failover feature is designed for a redundant configuration that contains a maximum of two devices in a device group, with one active traffic group.
An HA group is typically configured to fail over based on trunk health in particular. Trunk configurations are not synchronized between units, which means that the number of trunk members on the two units often differs whenever a trunk loses or gains members. The HA group feature makes it possible for failover to occur based on changes to trunk health instead of on system or VLAN failure.
Only one HA group can exist on the BIG-IP system. By default, the HA group feature is disabled.
To summarize, when you configure the HA group, the process of one BIG-IP device failing over to the other based on HA scores is noticeably faster than if failover occurs due to a hardware or daemon failure.
The BIG-IP® system calculates an HA score based on these criteria:
A weight is a health value that you assign to each object in the HA group (that is, pool, trunk, and cluster). The weight that you assign to each object must be in the range of 10 through 100.
The maximum overall score that the BIG-IP system can potentially calculate for a device is the sum of the individual weights for the HA group objects, plus the active bonus value. There is no limit to the sum of the object weights for the HA group as a whole.
For each object in an HA group, you can specify an optional setting known as a threshold. A threshold is a value that specifies the number of object members that must be available to prevent failover. If the number of available members is less than the threshold, the BIG-IP system assigns a score of 0 to the object, so that the score of that object no longer contributes to the overall score of the device.
For example, if a trunk in the HA group has four members and you specify a threshold value of 3, and the number of available trunk members falls to 2, then the trunk contributes a score of 0 to the total device score.
If the number of available object members equals or exceeds the threshold value, or you do not specify a threshold, the BIG-IP system calculates the score as described previously, by multiplying the percentage of available object members by the weight for each object and then adding the scores to determine the overall device score.
The threshold that you define for pools can be less than or equal to the number of members in the pool. For clusters, the threshold can be less than or equal to the number of possible blades in the chassis, and for trunks, the threshold can be less than or equal to the number of possible members in a trunk for that platform.
An active bonus is an amount that the BIG-IP system automatically adds to the overall score of the device running an active traffic group. An active bonus ensures that the device remains active when its score would otherwise temporarily fall below the score of the device running the standby traffic group. The active bonus that you configure can be in the range of 0 to 100.
A common reason to specify an active bonus is to prevent failover due to flapping, the condition where failover occurs frequently as a trunk member switches between availability and unavailability. In this case, you might want to prevent the HA scoring feature from triggering failover each time a trunk member is lost. You might also want to prevent the HA scoring feature from triggering failover when you make minor changes to the BIG-IP system configuration, such as adding or removing a trunk member.
Suppose that the HA group on each device contains a trunk with four members, and you assign a weight of 30 to each trunk. Without an active bonus defined, if the trunk on one device loses some number of members, failover occurs because the overall calculated score for that device becomes lower than that of a peer device. You can prevent this failover from occurring by specifying an active bonus value.
Although you specify an active bonus value on each device, the BIG-IP system uses the active bonus specified on the active device only, to contribute to the score of the active device. The BIG-IP system never uses the active bonus on the standby device to contribute to the score of the standby device.
To decide on an active bonus value, calculate the trunk score for some number of failed members (such as one of four members), and then specify an active bonus that results in a trunk score that is greater than or equal to the weight that you assigned to the trunk.
For example, if you assigned a weight of 30 to the trunk, and one of the four trunk members fails, the trunk score becomes 23 (75% of 30), putting the device at risk for failover. However, if you specified an active bonus of 7 or higher, failover would not actually occur, because a score of 7 or higher, when added to the score of 23, is greater than or equal to 30.