Figure 5 NLB cluster
NLB supports load sharing and redundancy among servers within a cluster. To implement fast failover,
NLB requires that the switch forwards network traffic to all servers or specified servers in the cluster, and
each server filters out unexpected traffic.
In NLB unicast mode, when a server joins the cluster or a failover occurs, a packet with a virtual source
MAC address is sent within the cluster. The switch then adds the virtual MAC address to its MAC address
table, and packets destined for the server use the virtual MAC address (although not used by the server)
as their destination address. If the virtual MAC address never ages out, the switch forwards packets only
through the port associated with the virtual MAC address rather than all ports connected to the servers
within the cluster.
To address this issue, disable MAC entry aging timer refresh based on destination MAC address to age
out the virtual MAC address, so that the switch can forward packets to all servers within the cluster.
Configuring the MAC learning limit on ports
To prevent the MAC address table from getting too large, you can limit the number of MAC addresses
that a port can learn.
To configure the MAC learning limit on a Layer 2 Ethernet interface or all ports in a port group:
Step
1.
Enter system view.
2.
Enter interface
view or port group
view.
3.
Configure the
MAC learning limit
on the interface or
port group.
Command
system-view
•
Enter Layer 2 Ethernet interface
view:
interface interface-type
interface-number
•
Enter port group view:
port-group manual
port-group-name
mac-address max-mac-count count
26
Remarks
N/A
Use either command.
Settings in Layer 2 Ethernet interface view
take effect on the interface only. Settings in
port group view take effect on all member
ports in the port group.
No MAC learning limit is configured by
default.
Layer 2 aggregate interfaces do not support
this command.