GMPLS unifies network layers

Generalized Multi-protocol Label Switching extends net intelligence from the network edge through the core and back to the edge under a unified control plane.

Carriers are seeking to gain efficiencies and develop advanced services by merging the worlds of IP and optical networking. However, first they need to overcome the extremely complex multilayer architecture that’s been cobbled together to carry IP service over networks designed to support voice and fixed circuits. Ultimately, they need one control plane that extends from IP at Layer 3 right down to the optical transport level at Layer 1.

Generalized Multi-protocol Label Switching (GMPLS) aims to meet this need by extending network intelligence from the network edge through the core and back to the edge under a unified control plane.

A proposed IETF standard, GMPLS is still in development and isn’t expected to be ready for wide-scale deployment for 1 to 2 years. But the technology isn’t entirely new, because it builds on many of the gains that came through the development and standardization of Multi-protocol Label Switching (MPLS ), which has simplified network architecture by replacing the need for ATM and frame relay equipment to oversee traffic engineering.

MPLS improves IP scalability and quality of service by creating virtual label-switched paths (LSP) across a network of label switching routers (LSR ). GMPLS’ primary enhancement to MPLS is its capability to establish connections at Layer 1.

GMPLS can be deployed in two ways: overlay model or peer model. In an overlay model, also called a UNI, the router is a client to the optical domain and interacts only with the directly adjacent optical node. In the overlay model, the actual physical light path is decided by the optical network and not by the router.

In the peer model, the IP/MPLS layer operates as a full peer of the optical transmission layer. Specifically, the IP routers are able to determine the entire path of the connection, including through the optical devices.

The aim of GMPLS – both peer and overlay models – is to extends the reach of MPLS from routers through to the optical domain, where forwarding decisions are based on time slots, wavelengths or physical ports (called “implicit labels” in GMPLS terminology), not packet boundaries. GMPLS enables such cross-domain peering by supporting new classes of LSRs, including dense wavelength division multiplexers, add/drop multiplexers and optical cross-connects.

The most significant aspect of GMPLS is the way it influences how labels are requested and distributed, bandwidth is allocated and network failures are “communicated.

GMPLS uses Interior Gateway Protocol (IGP) extensions to support various link types – normal, nonpacket and forwarding adjacencies into the link-state database. If nodes at both ends of the link can receive and transmit packets, GMPLS identifies them as a normal link. If not, they become a nonpacket link. If an LSR creates and maintains a label-switched path, it can announce the LSP into the IGP as a forwarding adjacency.

Crucial to this approach is GMPLS defining the hierarchy of LSPs. This lets the nesting of LSPs support the establishment of traffic trunks. The function is similar to MPLS’ support for label stacking, in which many smaller LSPs can be aggregated into one larger LSP. GMPLS operates much the same as LSPs operate, which is as a virtual representation of physical paths.

Under the hierarchy GMPLS sets up, LSPs that begin and end with packet-switched nodes are at the bottom, followed in ascending order by those tied to TDM switching nodes, lamba switching nodes and fiber switching nodes.

GMPLS is expected to help service providers dynamically provision bandwidth and capacity, improve network restoration capabilities, and reduce operating expenses. New revenue-generating services such as optical VPNs also might spring from GMPLS. Another anticipated gain comes from GMPLS supporting open standards. This will let carriers use best-of-breed equipment as they build out their networks.

Demand for GMPLS should grow as IP traffic and services increases. Yet challenges remain. Vendors need to establish business cases that support GMPLS’ introduction. And if companies are to gain maximum efficiency, they must overcome their own organizational roadblocks that separate optical transport and IP administrative domains.

Stewart is director of product strategy for Juniper. He can be reached at jstewart@ juniper.net.