Alteon Application Switch Operating System Application Guide
High Availability
518
Document
ID:
RDWR-ALOS-V2900_AG1302
In this example, VRRP is configured as active-active. Both Alteons are OSPF-enabled and receive
traffic. The following is a further explanation of
Figure 77 - OSPF VRRP Topology Using Cost
1. The cost of the first Alteon is less than the cost of the second Alteon.
2. Mobile clients send traffic from network 20.20.20.x through the first Alteon to GGSN on network
30.30.30.x.
3. Alteon intercepts and redirects the traffic based on the HTTP policy of the 10.10.11.x network.
4. The 10.10.10.x network does not appear in the OSPF routing and is accessed only by Alteon.
5. If the link between the first Alteon and the 10.10.11.x network fails, OSPF is not affected
because the interface of the 10.10.10.X network is not bound to OSPF.
6. The traffic passes from the mobile clients to the first Alteon and the service is interrupted.
7. If the link fails when the traffic returns from the servers in the 10.10.10.x network, traffic
returns through the second Alteon. This causes an asymmetric routing traffic flow.
VRRP cost update support does not require any changes to the OSPF settings. The VRRP
functionality is part of the existing tracking options. This enables OSPF to remain a pure routing
protocol regardless of the services running on top of it.
OSPF maintains a cost value per interface flexibility designed for routers creating deterministic
paths. In the example in
Figure 77 - OSPF VRRP Topology Using Cost Updating, page 517
, the
traffic flow is handled as a service with path dependencies. That is, the service paths are related and
affect one another.
You can set the OSPF cost increment for the VR (single interface), VR group (multiple interface), and
group (multiple interface). For more information on configuring the OSPF cost, refer to the Alteon
Application Switch Operating System Command Reference.
IPv6 VRRP Support
Alteon supports using IPv6 with VRRP. For background information on IPv6, see
This section describes the following topics:
•
IPv6 VRRP Support Overview, page 518
•
•
IPv6 VRRP Configuration, page 519
•
IPv6 VRRP Information, page 520
IPv6 VRRP Support Overview
IPv6 hosts on a VLAN usually learn about other routers by receiving IPv6 routing advertisements.
The routing advertisements are multicast periodically at a rate such that the hosts usually learn
about the other routers within a few minutes. They are not sent frequently enough for the hosts to
rely on them to detect router failures.
IPv6 hosts can also use the neighbor discovery mechanism to detect router failure by sending
unicast neighbor solicitation messages to the other routers. By using the default setting, it takes a
host about 38 seconds to learn that a router is unreachable before it switches to another router.
IPv6 VRRP support provides a much faster mechanism for the switch over to a backup router than
can be obtained using standard neighbor discovery procedures. Using IPv6 VRRP support, a backup
router can take responsibility for the virtual router master within seconds. This is done without any
interaction with the hosts, and a minimum amount of traffic in the subnet.
Two types of addresses are used in IPv6 that facilitate VRRP support:
•
Unicast address—The global unicast address is an address that is accessible and identifiable
globally.
