New network under construction

Telcos are well aware that end users' demands are changing. Customers want nontraditional services at the same quality level as voice, and they are starting to demand Internet and differentiated services, Internet protocol multicast, advanced, value-added IP services and telco-quality voice-over-IP capability.

Service providers are responding by improving their networks with high-performance switches and routers. But as they add the hardware, questions arise. Will the equipment really provide the high performance needed? Will the network be scalable as business grows and more services are offered? Are the network components standards-compliant and interoperable? Can the value-added service be delivered through a virtual private network (VPN)?

Major networking companies are acting to ensure that service providers can offer more services to more customers over an open, scalable, high-performance network. Since late 1996, equipment vendors have been working with service providers under the auspices of the Internet Engineering Task Force to create a standard for multiprotocol label switching (MPLS).

Alcatel, Bay Networks, Ericsson, General DataComm, IBM, Toshiba and Ascend Communications have announced plans to support Cisco Systems' proprietary tag switching and MPLS in their products. Companies such as Juniper and Ennovate also will support the capabilities.

At its core, MPLS is a multilayer switching technology that provides the Layer 2 speed of switches and the Layer 3 intelligence of routers. End users benefit from the integration by gaining advanced features such as quality of service (QOS) and reliable traffic engineering throughout the network.

The technology is designed to boost scalability, reliability and value-added service delivery for those capabilities where service providers need it most: at the large-scale WAN level.

MPLS operates by assigning labels to multiprotocol frames so they can be transported across packet or cell-based networks. It is based on the concept of label swapping, in which units of data-a packet or a cell-carry a short, fixed-length label that tells switching nodes how to process the data. When the information reaches the designated location, the label is removed. Through this process, data delivery is assured without sacrificing speed.

So far, the MPLS working group is finishing the final details of MPLS. If all goes according to plan, the first phase of the standard will be ready by late this year or early 1999 (Figure 1).

Cisco's tag switching formed much of the basis for the MPLS standard, along with Aris from IBM, CSR from Toshiba and, to a lesser extent, Ipsilon's IP switching.

Like MPLS, tag switching uses tags-the same as MPLS labels-to guide data to its proper end. Tag edge routers and tag switches use standard routing protocols-open shortest plan first, intermediate system to intermediate system, and enhanced interior gateway routing protocol-to identify routes through the network. These interoperate with non-tag switching routers.

The routers and switches use the tables that the standard routing protocols generate to assign and distribute tag information via the tag distribution protocol. Tag routers receive the tag distribution protocol information and build a forwarding database that makes use of the tags.

When a tag edge router receives a packet for forwarding across the tag network, it analyzes the network layer header, performs applicable network layer services, selects a route for the packet from its routing tables, applies a tag and forwards the packet to the next hop tag switch.

The tag switch receives the tagged packet and switches it based on the tag without re-analyzing the network layer header. Once the packet reaches the tag edge router at the network's egress, the tag is stripped off and the packet is delivered (Figure 2).

Cisco's tag switching software will be updated to comply with the MPLS standard when it is completed.

Three distinguishing features Tag switching allows for three features that service providers require: integration of IP and asynchronous transfer mode, explicit routing and VPNs.

Support for IP QOS and IP multicast across an ATM backbone requires a complex translation or mapping between the connectionless IP and connection-oriented ATM protocols. Tag switching eliminates the complexity by enabling the ATM switches to support advanced IP services and protocols directly, reducing operational costs and bandwidth and improving time-to-market for new services.

Tag switching ensures easy scalability by integrating ATM with Layer 3 routing. It enables ATM switches to be integrated fully into Internet core networks, including border gateway protocols without the scalability problems of a pure Layer 2 ATM network ringed by a router overlay using permanent virtual circuits (PVCs) or switched virtual circuits.

Tag switching further enhances scalability in an ATM network by reducing the number of tags needed through a capability called virtual circuit merge, which is similar to the concept of a multipoint-to-point virtual circuit. And it has the architectural flexibility to support ATM, frame relay and tag switching/IP services on the same high-performance ATM switched infrastructure.

Tag switching also allows explicit routing, which enables enhanced traffic management, load balancing across the core and cost savings by using the installed wiring and networking hardware base.

Explicit routing specifies a path through the network and selects a set of packets to follow that path. In this way, forwarding decisions are based on criteria other than destination addresses. In other words, explicit routing enhances traffic management in router-based internetworks by integrating Layer 2 circuit capabilities.

Explicit routing also is important when fiber is not available along the optimum traffic path, when the cost of bandwidth may result in high link use on some paths, when a link failure may cause other links to become overloaded and when links can be overloaded by unexpected changes in traffic patterns.

Explicit routing lets the service provider engineer the flow of packets to cope with these situations. Tag switching extends these capabilities to internetworks that are built completely from Layer 3 routers.

VPNs for data between enterprise sites are built from frame relay PVCs across a service provider's frame relay network. But because of the growth of intranets within enterprises, traffic patterns are shifting from the traditional hub-and-spoke-for which frame relay is a good match-to a meshed, any-to-any pattern that does not scale well with connection-oriented access services.

Service providers also encounter scalability limits in providing frame relay-based managed router service offerings. Tag switching supports IP-based VPNs-a potential source of major revenue growth-on either a frame relay/ATM backbone or a gigabit router backbone.

By tagging packets based on both the destination address and the VPN to which the packet belongs, tag switching-enabled products can accommodate large numbers of IP VPNs with overlapping IP address spaces and QOS-based service level agreements. And it can provide the support with the same data security as a Layer 2 frame relay network (Figure 3).

The next generation of labeling technology should provide an infrastructure robust enough to handle services that aren't even on the radar screen, and do so cost-efficiently.