Znyx Networks bh5700 Manuale Utente
VRRP
Since most end nodes use default router addresses, the change of the default router address during
a switch failover would require the end nodes to reconfigure. Layer 3 switches that failover must
maintain the default router address to maintain the end node's IP transparent failover. The Virtual
Router Redundancy Protocol (VRRP, RFC 2338) running in the Surviving Partner switches
provides transparent movement of the default router address. VRRP maintains the notion of a
Master switch and one or more Backup switches. This group of switches presents a virtual router
IP address that can be used by hosts on that net as their default route.
a switch failover would require the end nodes to reconfigure. Layer 3 switches that failover must
maintain the default router address to maintain the end node's IP transparent failover. The Virtual
Router Redundancy Protocol (VRRP, RFC 2338) running in the Surviving Partner switches
provides transparent movement of the default router address. VRRP maintains the notion of a
Master switch and one or more Backup switches. This group of switches presents a virtual router
IP address that can be used by hosts on that net as their default route.
If a Backup switch determines the Master switch is no longer available, one of the Backup
switches will assume the role as Master. Physically, each switch maintains a link to the local
network. Only the Master switch answers to the default gateway, and the hosts on that net have
no need to relearn the router address.
switches will assume the role as Master. Physically, each switch maintains a link to the local
network. Only the Master switch answers to the default gateway, and the hosts on that net have
no need to relearn the router address.
In an HA configuration, the goal is to avoid any single point of failure. VRRP provides a good
mechanism to provide a static route for a local network, but a true HA configuration must also
provide redundant connections for the host. Providing a virtual router for the local network is not
enough. Take the simple case of two hosts on the local network with a connection to the virtual
router. Each host needs a connection to each physical switch participating in VRRP. In the
simplest configuration, each host would have one connection to the network. An HA solution
would include redundant connections from each host to each switch in the virtual router.
mechanism to provide a static route for a local network, but a true HA configuration must also
provide redundant connections for the host. Providing a virtual router for the local network is not
enough. Take the simple case of two hosts on the local network with a connection to the virtual
router. Each host needs a connection to each physical switch participating in VRRP. In the
simplest configuration, each host would have one connection to the network. An HA solution
would include redundant connections from each host to each switch in the virtual router.
Combining the features of Surviving Partner on the switches and HA bonding drivers on the hosts
allows implementation of this true HA configuration.
allows implementation of this true HA configuration.
zlmd
In addition to complete switch failover, single link failure must be properly handled. The Link
Monitor Daemon zlmd, monitors the link status of each port. If a link goes down, zlmd
communicates with the VRRP daemon (vrrpd) to change its priority. Changing the VRRP
priority results in movement of switching functionality. By combining zlmd with the zlc
application, links connected to hosts that have not failed can be deterministically moved to the
new master switch if desired. Supported modes include:
Monitor Daemon zlmd, monitors the link status of each port. If a link goes down, zlmd
communicates with the VRRP daemon (vrrpd) to change its priority. Changing the VRRP
priority results in movement of switching functionality. By combining zlmd with the zlc
application, links connected to hosts that have not failed can be deterministically moved to the
new master switch if desired. Supported modes include:
•
switch - The switch with the greatest number of UP links becomes the Master for all
VLANs under HA management.
VLANs under HA management.
•
Vlan - The switch with the greatest number of UP links in that particular VLAN becomes
the Master for that particular VLAN. If the switch has additional VLANs, they each
change independently.
the Master for that particular VLAN. If the switch has additional VLANs, they each
change independently.
•
Port - The Master will remain the Master for that particular VLAN until all ports in that
VLAN are down. The Backup then becomes the new Master for that VLAN. Failed links
move their connectivity through the Backup Switch and the switch interconnect to reach
the Master Switch. This option alleviates the need to move all nodes to a new switch just
because a single link goes down.
VLAN are down. The Backup then becomes the new Master for that VLAN. Failed links
move their connectivity through the Backup Switch and the switch interconnect to reach
the Master Switch. This option alleviates the need to move all nodes to a new switch just
because a single link goes down.
NOTE: All modes require inclusion of the interconnect in the VLAN. The ISL
connection between the two Base switches is port 23 for the Ethernet Switch Blade. The
ISL connection between the two fabric slots in port 51.
connection between the two Base switches is port 23 for the Ethernet Switch Blade. The
ISL connection between the two fabric slots in port 51.
Ethernet Switch Blade User's Guide
release 3.2.2j
page 28