If you were configuring routers in the early 1990s, you remember that they had to support about 10 different internetworking protocols. Most of them have gone the way of the dinosaur, and almost all data communications now use IP, which significantly reduces cost and complexity.
A decade ago, MPLS emerged as another unifying technology. Forwarding most packets at Layer 2, MPLS allowed service providers to build large-scale IP-based networks that supported traffic engineering, connection-oriented packet transport and VPNs, features that were hard or impossible to implement with native IP.
The widespread deployment of MPLS and Carrier Ethernet caused transmission price-per-bit to drop by orders of magnitude and put frame relay and asynchronous transfer mode (ATM) on the endangered technologies list.
Encouraged by the rapid uptake and enormous success of MPLS in the IP world, vendors tried to apply the same principles to transport and optical networks. They tried to unify SDH, OTN, and DWDM, and developed Generic MPLS (GMPLS), which so far has been a failure. The mentality of transport network operators is obviously incompatible with the connectionless unpredictable self-adjusting world of IP.
The ITU's standardization sector, ITU-T, decided in 2006 to merge the same architectural principles used in transport network technologies like SDH, SONET and OTN with MPLS. The ITU tried to recycle GMPLS into its own MPLS-like technology called Transport-MPLS (T-MPLS).
The first result of the joint effort had a list of 115 requirements that identified MPLS-TP-specific requirements in six major areas. Some of these requirements specify the mandatory or recommended use of existing MPLS technologies or components, but many require new functionality and significant reworking of existing MPLS control and management protocols (see below).