Glen, Here's a list of technical issues for whatever it is worth:
Transport networks assume a service to be a circuit with two end-points which optionally can be protected by some transport mechanisms. Each of such services is unrelated to others. In a routed network a transport service corresponds to a link adjacency between routers. To design a resilient router network it is desired that links between different pairs of routers are unrelated to each other and don't fail together (are shared-risk-link-group diverse). Hence links are related in being unrelated. Now let's start: 1) transport folks don't disclose their physical topology: no chance to design a reliably resilient network on top. Workaround: ask for protected transport links. This increased availability comes with the cost of duplicating transport resources. Actually it is often more than double since the protecting circuit is a bit longer than the protected one. 2) transport folks do disclose the physical topology including nodes. Considering those nodes as single point of failure, it requires more BW available in Routers to cover failures and tricky configuration. 3) alternatively you could ask to install more transport equipment to avoid SPOF which obviously comes with a price tag too 4) once settled, the Transport network may grow. This means that circuits once in a while get moved to other resources. Typically the result of a transport network re-planning, a protection action or simply a forgotten reversion after a maintenance event. Consequently, whatever was achieved in 2) or 3) may just become void over time. There are basically two ways to deal with the complexity: 1) create an information exchange that let networking and transport controllers take informed decisions: You may want to look at https://datatracker.ietf.org/doc/draft-ietf-teas-interconnected-te-info-exchange/ 2) forget about switching anything in Transport, nail the structure down in config and move on. You may like then https://meetings.ripe.net/see3/files/Ian_Farrer-The_Terastream_Native_IPv6_Network_Architecture.pdf If you now think transport is all the same, irrespective of technology, then I suggest to take another breath: 1) since multiplexing lambdas is passive, Operators often do not count WDM or ROADM nodes as SPOF. Note: multiplexing is indeed passive but followed by Amplifiers which are not. 2) in SONET one could connect any vendor's node to everyone else's e.g. With OC48. This is not true in DWDM. While the Ethernet connecting to a wavelength is a standard, the wavelength itself is proprietary and linked to the component vendor of the transceiver. Transponders are in fact acting as adaptor plugs between standard and proprietary transport. As a result, if you find a left-over Gen-1 transponder and want to hook it up to a Gen-3 transponder of the same system vendor - it may not work because the component supplier changed. It also means you can't simply re-structure a DWDM network in an on-demand manner if such effects are present. I leave it up to you to draw your own conclusions. Gert Sent from my Apple ][)
