wilamowski-b-m-irwin-j-d-industrial-communication-systems-2011

Quality of Service 19-13
2 and 3 will start to drop, but WRED will also continue to randomly drop packets with CoS 0 and 1.
The final result is that there will be more packets dropped with CoS 0 and 1 than with CoS 2 and 3.
19.4.2 High Availability Solutions for the Routers
As mentioned previously, another feature for the “service level” solution is the availability of the
network. None of the previous QoS solutions will work if the network devices (switches and/or routers)
where must be implemented are not working. This means that any complete network QoS solution must
take into account the redundancy for, at least, the most sensible routers in the network.
Nowadays, this is implemented almost universally by using the Virtual Router Redundancy Protocol
(VRRP). The VRRP [26] obliges to duplicate routers and links to ensure continuity of service across
failures. It introduces the concept of “virtual router” that is addressed by IP clients requiring routing
service. The real routing service is provided by physical routers running the VRRP.
There are a number of descriptive terms introduced by VRRP:
• Virtual router, a single router image created through the operation of one or more routers running
VRRP.
• Virtual router ID or VRID, a numerical identification of a particular virtual router. It must be
unique on a network segment.
• Virtual router IP, an IP address associated with a VRID that the other hosts use to obtain network
service. It is managed by the VRRP instances belonging to a VRID.
• Virtual MAC address, a predefined MAC address used for all VRRP actions (for Ethernet) instead
of the real adapter MAC address(es). It is derived from the VRID.
• Master, the VRRP instance that performs the routing function for the virtual router at a given
time. Only one master is active at a time for a given VRID.
• Backup, other VRRP instances for a VRID that are active, but not in the master state. Backups are
ready to take on the role of the master if the current master fails.
• Priority, a value assigned to different VRRP instances, as a way to determine which router will
take on the role of the master if the current master fails.
Figure 19.8 shows a simple VRRP configuration, with two routers connecting to a network cloud and
VRRP providing a resilient routing function for the client machines in the local area network.
Router rA is the master of virtual router VRID 1 and router rB is at the backup state. When a VRRP
instance is in the master state for a VRID, it sends multicast packets to the registered VRRP multicast
address advising other VRRP instances that it is the master for the VRID, with a twofold objective:
• If a VRRP instance with a higher priority for that VRID is started, the new VRRP instance can
force and election and take on the master role.
• VRRP instances in the backup state for the VRID listen for the master’s packets; if an interval
elapses without a packet being received, the instances in the backup state take action to elect a
new master.
In Figure 19.8, one of the following must be true about router rA:
• It has a higher priority than router rB.
• If both routers have the same priority, its interface IP address is higher than that of router rB.
If router rA fails, after a short interval router rB would notice that no multicast packet has been
received and transitions to the master state, taking over the handling of the virtual IP address and sending
its own multicast packets.
The time router rB waits before making its state transition is called the master down interval and is
based on the length of time between master updates (advertisement interval) and a value called skew
time calculated from the priority value.
© 2011 by Taylor and Francis Group, LLC