Just my opinion - I'm sure others will have different ideas.
You are always going to be fighting against 3 parameters pulling your design in different directions. Idealy you want a design that is:
Strong
Light
Cheap
However, resilient and agile desigs are more expensive. Cheap, agile designs lack high-availability. Cheap resilient designs lack flexibility. So it is always going to be a compromise.
If resilience is key for users connecting at the edge workstations they are always going to be exposed to a single point of failure wherever their ethernet cable plugs in. How about shifting everyone to wireless and flooding the environment with high throughput resilient "ac" wireless from say Aruba IAP series.
Backhaul these into, say, 5400Rzl switches with full routing stack and redundant power supplies and management cards.
That's going to be a fairly "bomb proof" topology. What are the users connecting to - can you put resiliece around the entire stack out to the internet edge (resilient WAN routing) and beyond (different circuits, different Service providers, BGP routing etc)?
Lots to think about - try and "stand further back" - draw the traffic flow end to end, workstation all the way through to server or cloud service and see what questions you can ask yourself? How would this survive a failure of component XXX?
Good luck - let us know how you get on
Ian
Hope that helps - please click "Thumbs up" for Kudos if it does ## ---------------------------------------------------------------------------## Which is the only cheese that is made backwards? Edam! Tweets: @2techie4me
