Discussion on the link route of data layer traffic between VLANs of VPC

一页书

As shown in the figure: Peer KeepAliver link is the Channel of E4/18-19 (I am too lazy to write on the right)
          Peer link is the channel assigned to E3/19-20
          E4/22 (left)-E4/28 (right) CISCO and the leader recommend independent lines to run the three-layer routing

Peer KeepAliver link is responsible for routing links, carrying heartbeat data packets, and performing double-Active detection.

But in the video explanation, it highlights the state detection of Peer link

When the Peer link is normal, this guy is a soy saucer.

The traffic undertaken by the Peer link is mostly control plane traffic, which only forwards some data when the link fails.

(What kind of data is it? The video is directly bludgeoned)

Okay, here comes the problem.

Where does the traffic of the data plane go out, in essence, where does the inter-VLAN routing traffic go?

If you go from the Peer KeepAliver link of the M card, you are blind, and it is to maintain the status detection of the Peer link.


What's more, in the recommended setting, it is also thrown into the VRF, which is an independent existence.

If it’s the Peer link of the F card, let’s not talk about Peer’s anti-loop mechanism. In the video, the F132XP card is used. Can God tell me that F1 can still run Layer 3 traffic?

But HSRP and TRUNK are set here, this is understandable, because they are all serving Peer Link

According to the document, the Peer link carries traffic such as vPC and non-vPC VLAN, HSRP Hellos, etc., then where did the data layer traffic go? Your sister's

As mentioned earlier, if two N7Ks only have one M card each, when the M card is hung on one end, it will not be released.

The lower VLAN routing passes through the Peer link via the HSRP protocol.

The anti-loop coming from the Peer link port will no longer be put into the Peer link port, so it is recommended to configure two M cards at one end.


After going around for so long, where did the damn VLAN data traffic go? Control flow and data flow have been separated? ? ?

to sum up:

Judging from the redundancy of the M card, the VLAN traffic must fly to this old boy, but the M card is not for you to run TRUK! ! !


How to check your VLAN tag?


From the HSRP point of view, the deceptive Peer link, your uncle is able to take off your clothes and put on clothes, but you don’t allow you to stop by (Peer port is anti-ring, uncle’s)

From vPC peers For neighboring devices connected to vPC:

Appears as a single exchange, which is to logically see the two damn N7Ks above as "Uncle Siamese"

Then can I logically understand your "uncle" as:

You connect the HSRP control traffic of the Peer link and the independent line of the M card through the routing protocol

(Such as OSPF in the video) data traffic


Integrate into the new routing protocol you agreed to. What protocol? As long as you have just finished the QJ birth, you can know what it is called.

In this way, you not only get through the control layer, but also get the data layer.

I don't know how you guys think about the direction of routing traffic between VLANs?

I personally think that it is still a logical "split", first regarded as two wastes, and finally combined into "Transformers"

I can only comfort myself so much. Only in this way can I understand why VLANs can be routed.