I just did a ton of reading about the different multi-tier network topology options as outlined by Cisco, and now that I'm looking at the physical options (End of Row (EoR) vs Top of Rack(ToR)), I find myself confused about how these fit into the logical constructs.

With ToR it also maps 1:1: at the top of each rack there is a switch(es) that essentially act as the access layer. They connect via fiber to other switches, maybe chassis-based, that act as the aggregation layer, that then connect to the core layer.

With EoR it seems that the servers are connecting directly to the aggregation layer, skipping the access layer all together, by plugging directly into what are typically chassis switches. In EoR then is the standard 3-tier model now a 2-tier model: the servers go to the chassis switch which goes straight to the core switch?

The reason it matters to me is that my understanding was that the 3-tier model was more desirable due to less complexity. The agg switch pair acts as default gateway and does routing; if you use up all of your ports in your agg layer pair it's much more complicated to add additional switches, than simply adding more switches at the access layer. Are there other downsides to this layout? Does this 3-tier architecture still apply in some way in EoR?



ToR vs EoR (phyiscal) doesn't dictate logical design. The EoR switches could just as easily be the access layer (or L2 switches as you implied). Regardless of ToR or EoR, your logical design could even make the access layer L3. It's all about how you want to logically break up the network and where you have L2 adjacency requirements.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.