Re: As long as there is one cable in the universe...
"Especially now Ethernet's escaped from the WAN and spanning tree problems can go global in a hurry if you're not careful."
"The answer my friend, is blowing in the TRILL
and your router will soon be in the bin"
https://tools.ietf.org/html/draft-ietf-trill-irb-03
Seriously, routers end up being a major bottleneck and this development solves that plus a big mess of wiring.
TRILL itself was designed with widescale deployments in mind, specifically _because_ of spanning trees storms and convergence issues. Spanning tree was never envisaged as being more than 8 hops wide.
The other good thing about TRILL is that you don't have to faff about with LACP. As with fibre channel networks, when you add more links between switches, they "just work" (bear in mind that every LACP disturbance == a spanning tree rebuild, even if the LACP is to an edge device.)
If you can wean yourself off Cisco, there are a bunch of devices out there from other makers all based on the Broadcom Trident2 chipset and all selling for about 1/3 of cisco's prices (and I don't mean their "list" prices) for 64*10GB/s(*). The big hassle at the moment is that they tend to be "all SFP" or "all copper", which doesn't suit campus distribution well. Terbabit L3 forwarding rates aren't to be sneezed at.
(*) usually 48*10Gb + 4*40Gb, but there are QSFP-only versions and a few 100Gb/s versions too. Higher speeds allows rack switches == less cable == less mess - and that keeps us on topic :)