I was brought in to perform an assessment of a customers network and for the most part the environment is in a decent state with infrastructure, redundant WAN, HA , L7 etc. But, I've come across some weird network design decisions at their offices and the response I've received when I asked about it was that this was intentional. The layout is below:
L3 Core - 2 stacked L3 capable switches with SVIs. All intervlan routing between occurs here and only traffic for the WAN or remote offices is routed to the edge.
All other switches in the building are stacked in sizes of 4-7 switches. They are L2 access switches for users, APs, printers, etc to connect into.
Below is a short snippet of the design that threw me:
L3 Core Stack -> A single 10G fiber run Switch Stack 1
Switch Stack 1 -> 2 10G fiber runs to Switch Stack 2
Switch Stack 2 -> 2 10G fiber runs the L3 Core Stack.
From my perspective, the 2 10G fiber connections between Switch Stack 2 and the L3 Core Stack should be in a LAG. Ideally a second 10G fiber run would be implemented from the L3 Core Stack to Switch Stack 1 and that would also be in a LAG. We would decommission the 2 10G fiber runs between Switch Stack 1 and Switch Stack 2.
Running L3 between all the Stacks and the Core isn't an option with the current infrastructure.
The pushback that I am receiving with this is making me question my network sanity. I see this as one spanning-tree blip away from a network crashing broadcast storm but maybe I'm wrong.
I'd appreciate any other points of view on this.
** Added/reworded for clarification.
No comments:
Post a Comment