VXLAN on a Collapsed Core/aggregation-layer design.

[KOAVA]

Hi all,

I would like your opinion and guidance for a network design concerning two separate Data Centers running on a Nexus 9000 Compact aggregation-layer design where i would like to have a DCI/Lan extension capabilities.

So, I would like to ask if it is possible to have a LAN extension solution between two Data centers using VXLAN when the network design is based on Nexus 9000 Compact Collapsed Core/aggregation-layer design.
For example, on each site there will be a Pair of Core/Aggregation 9K's running VPC to the Downstream TOR 9K's access layer switches where the servers will be Dual home attached with VPC.
Some VLANs/SVI's will be terminated on an FTD appliance that will be dual attached on the Collapsed Core 9K and some other will be locally connected on the 9K Core/Aggr device.

Is there any how to guide or a CVD for the above scenario?

Is this a valid go to solution and design or should i consider the Spine/Leaf topology? The server racks are less than 10 on each site so i want to keep it as simple as possible.

Thanks.

[Pavel Tarakanov]

If your purpose is to use VXLAN only between data centers, then you can build back-to-back VPC BGW topology and enable VXLAN only on them. In such case, core switches will also perform BGW role.

https://www.cisco.com/c/en/us/products/collateral/switches/nexus-9000-series-switches/white-paper-c11-739942.html

[KOAVA]

Yes my main goal is to have L2 extension between our 2 Data Centers. There is no need to run vxlan inside the DC because there are not too many Racks/Servers. The collapsed core design works just fine for us.

Thanks for the info, I will have a look at it and I'll come back for any questions.

[KOAVA]

If i understand correct, in the   BGW back to back topology i will have to physically interconnect the BGWs. This creates physical-connectivity issues because i already have a layer 3 core between the two DC's which i want to take advantage of. Due to high telecom costs, there is no option of buying more circuits from the local providers for the back-to-back connectivity of the BGWs'.

So in that case i understand that theBGW-to-cloud model will be a better choice.

But one thing that i should mention,is that the is that on one of the two DC's, the layer 3 core link will be terminated on the Collapsed Core/aggregation 9K switch which will also be the BGW.

The other side of the link will be terminated on a dedicate layer 3 router which wil have layer 3 links back to the BGW.

Do you see any stoppers issue here?

Thanks.

[KOAVA]

I have also checked the VXLAN EVPN Multi-Site with vPC BGWs whitepaper but again it mentions that the the BGWs should be interconnected to each other.

https://www.cisco.com/c/en/us/products/collateral/switches/nexus-9000-series-switches/whitepaper-c11-742114.html

Actually this topology is pretty close to what i want achieve with the difference that the layer 3 link will be terminated to the Collapsed Core/aggregation 9K switch and not back to back to BGWs. Is it valid?(Something like the OTV on a stick topology)

Thanks

[KOAVA]

@Pavel TarakanovSorry i confused. While i was reading the whitepapers i thought that the BGWs and the core switches should be on separate machines? But you wrote that,  core switches will also perform BGW role.

If that is possible, then i suppose is valid to Use the same chassis as a BGW and Core/Aggr Switch  and then use vPC BGW nodes to connect multiple legacy data center sites.

Thanks