Specifications

50 Barracuda Load Balancer Administrator’s Guide
Before joining two systems together, each Barracuda Load Balancer must meet the following
requirements:
• Barracuda Load Balancer models 340 or higher.
• Be the same model as the other Barracuda Load Balancer and on the same version of firmware.
• Be able to access all Real Servers.
• Be installed on a unique management IP address. The Barracuda Load Balancers use the
management IP address (over SNMP ports) to communicate for high availability.
• Be able to ping each other on the WAN interface (i.e. no firewall between them).
• The WAN interface on both Barracuda Load Balancers must be on the same switch (or physical
network).
Operation of HA
The two Barracuda Load Balancers that you use for HA make up a cluster. When you create a cluster,
the configuration is copied from the primary to the backup system. The primary system performs the
load-balancing. The other waits in standby mode and polls the primary Barracuda Load Balancer to
ensure that it is operational. If the backup system determines that the primary is no longer operational,
it takes over the VIP addresses and starts load-balancing just as the primary had been doing.
The backup Barracuda Load Balancer does not do any load-balancing or monitoring of Services or
Real Servers unless the primary system fails. Because it does not do any health monitoring while in
standby mode, all of the Services and Real Servers on a page such as
Basic > Services will have red
health indicators.
In a Route-Path deployment, the backup Barracuda Load Balancer automatically becomes active and
takes over operations if it does not receive a reply to its poll of the primary system within 5 seconds.
In a Bridge-Path deployment, the backup Barracuda Load Balancer automatically becomes active if
the backup system does not receive a reply to the poll within 30 seconds.
If the primary system fails and the network is caching ARP requests, then the network will not
immediately associate the MAC address of the backup Barracuda Load Balancer with the VIP
address. Because of this, the length of time it takes for the switchover from primary to backup is:
(the lifetime of the ARP cache)
+ (the time it takes for the backup Barracuda Load Balancer to determine
that the primary system is inactive)
For example, in a Route-Path deployment, if the ARP cache lifetime is 10 seconds, the switchover
will take about 5 + 10 = 15 seconds.
Recovery of the Primary System
If, after an outage, the primary system becomes available again, the backup will detect that the
primary is operational and stop load-balancing. The primary will take over the Virtual IP addresses
and the load-balancing. This switchover from backup to primary is accomplished within a few
seconds. Load-balancing will be disrupted only briefly.
Creating a Cluster and Removing the Cluster
To create a cluster of two Barracuda Load Balancers:
1. Complete the installation process for each system as described in Chapter 3 Initial Setup. For
Route-Path only, leave the LAN IP field blank on the backup Barracuda Load Balancer. To
verify this, go to the
Basic > IP Configuration page on the backup Barracuda Load Balancer.