Ethernet or Oc3 Bandwidth - What Do You Choose?

Is Ethernet or OC3 bandwidth the right choice for a network solution for your business? How would you make that decision?

If you are talking about a network operator / service provider and assuming your links are above 1 Gbps .... most likely you will need both OCx/EoS and Ethernet. After all you need to interconnect with other networks. If you have links below 1 Gbps at the access or core sides of the network then you would have to use OCx, most likely. Although some people tend to say that for a new network you can do whatever you want, reality is another thing.

OCx/STM-N are known for their reliability and predictability and still have quite a good precense in long haul. These are migrating to EoS (VCAT, LCAS, GFP) and OTN but they won't satrt to become "pure" Ethernet until 43G and supposedly there won't be any difference at 100G so Ethernet could prevail adopting some of the reliability factors and DWDM compatibility from OTN. There is still long ways to go before the true convergence. In fact IEEE HSSG and ITU-T are working together to assure neither one reinvents each other's wheels. Perhaps, if they succed, we would see single physical interfaces that could handle both: Ethernet for end-to-end connectivity and OTN for transport applications.

If you are thinking forward, perhaps you need to add OTN and EoS to your transport/core infrastrucure so it is ready for the future and yet flexible to interconnect to other peers using OCx, or deliver services below 1 GigE using EoS. People claiming they get "pure Ethernet" from their providers may not be paying attention to the EoS add/drop box that was delivered with the service contract and it is in fact receiving packets and shooting OCx frames back to the central office, or it arrived at a near aggregation location via OCx/EoS and then radioes to the customer premisses (perhaps over an internal OCx link too).

Below 1 GigE rates (I believe) there is no good way to deliver Ethernet directly to the customer (access network). it always has to be wrapped on some sort of (unreliable) transport scheme like VDSL, SHDSL, DOCSIS,... This is (in my opinion) what's keeping SONET/SDH alive. Once we all have fiber and beyond 1GigE then there is no issue. You would be able to go Ethernet/OTN all the way.

While Ethernet is becoming quite widespread and cost effective, often times the A and Z locations may require a decision in favor of OC3. SONET being readily available, it may be more cost effective (yes - even including the WAN module for OC3) in some locations. For the same A and Z locations, if both Ethernet and OC3 are available, I would pick Ethernet due to the ease of integration as long as the cost is within ~ 65% of the OC3 circuit.

OCx has history, resilience, and capacity, and is subject to suppliers wringing every last cent out of it as an asset given the emergence of long(er) distance Ethernet.

"Choose wisely", says the Templar Knight in Indiana Jones and the Final Crusade... likewise here. Unless you are committed to a long-term contract on OCx, don't extend it unless your business and application requirements say you won't need Metro-E flexibility until the end of the extension period. Even then, negotiate for huge concessions and/or cost-free upgrade to Metro-E at the end. As cost competitiveness of a service is based on the amount of market share: Metro-e /10GB-e will dwarf OCx over time.

Lastly, choose OCx/Metro-E service based on achievable and measurable SLAs first ..... then nail the supplier to the floor on price (as you do .... or .... you should do).