Руководство По Проектированию для Cisco Cisco Nexus 5010 Switch

Скачать
Страница из 28
 
Design Guide 
 
© 2010 Cisco Systems, Inc. All rights reserved. This document is Cisco Public Information. 
Page 21 of 28
 
You can also script this as follows: 
7k-1(config)# cli alias name vpcpreempt conf t ; vpc domain <domain-id> ; 
role priority 32767 ; int <peer-link> ; shut ; no sh * 
vPC Peer Link 
A PortChannel connects agg1 with agg2 and carries all access VLANs (defined by the user). This link also carries 
additional traffic that the user does not need to define, more specifically BPDUs and HSRP hellos, and MAC address 
synchronization between the vPC peers. This link is called peer link
On the Cisco Nexus 7000 Series, this PortChannel should be configured on dedicated-mode 10 Gigabit Ethernet 
interfaces across two different 10 Gigabit Ethernet line cards
This is by far the most important component in the vPC system, in that its failure, while not disruptive to existing vPC 
flows, may impair the establishment of new flows and isolate orphan ports. Configuring the peer link in a redundant 
fashion ensures virtually uninterrupted connectivity between the vPC peers. The following configuration illustrates 
how to configure the peer-link, which in this case is PortChannel 10. 
agg(config)# interface port-channel10 
agg(config-if)#  vpc peer-link 
agg(config-if)#  switchport trunk allowed vLAN <all access vLANs> 
vPC VLANs and non-vPC VLANs 
The PortChannel connecting the vPC peers should carry all the VLANs used by the vPC member ports. 
It’s also possible to carry the VLAN used by orphaned ports on this same link with the same caveat as a regular non-
vPC topology, which is, that upon losing the peer-link, communication between orphaned ports is interrupted (split 
subnet). If you want to avoid this problem you should make sure that servers are dual connected with a PortChannel 
to vPC ports. 
Alternatively, if you want to decouple vPC and non-vPC failure scenarios, you can use different VLANs for vPC-
connected devices and single-port attached devices (orphaned ports), and put the non-vPC VLANs and the peer link 
on different trunks. 
vPC Peer Keepalive  
Finally, a dual-active detection configuration needs to be put in place. The keepalive that resolves dual-active 
scenarios should never be carried as a VLAN on the peer link. Instead, it can be carried over a routed infrastructure, 
and it doesn’t need to be a direct point-to-point link. 
The following configuration illustrates the use of a dedicated Gigabit Ethernet interface for this purpose: 
vrf context vpc-keepalive 
 
interface Ethernet8/16 
  description tc-nexus7k02-vdc2 - vPC Heartbeat Link 
  vrf member vpc-keepalive 
  ip address 192.168.1.1/24 
  no shutdown 
 
vpc domain 1 
peer-keepalive destination 192.168.1.2 source 192.168.1.1 vrf vpc-keepalive