How Multiprotocol Label Switching Will Benefit the Internet
- What Is MPLS?
- Why Is MPLS Needed?
- How Is MPLS Done?
- When and Where Is MPLS Used?
- Who Is Doing MPLS?
- The Label Switching Paradigm
- A Quick Introduction to MPLS
- Evolution of Internet Network Models
- Basics of the Internet
- Internetworking Technology Basics
- More Basics: Graph Theory and Modeling Language
- The Promise of MPLS
- The Promise of the Promise of MPLS
Do not go where the path may lead, go instead where there is no path and leave a trail.A
Ralph Waldo Emerson
"It's the bandwidth, stupid."
This flip response from your favorite service provider after all of your keyboard pounding, all that excessive mouse clicking, and all of your "What's taking so long?" whining hit close to the mark in the past. The lack of bandwidththat is, the carrying capacity of the communications channelwas the primary culprit for long delays, low throughput, and even poor reliability problems. The constantly increasing demands on the Internet and its services require planning and technological innovation for its undeniable success to continue and grow. Indeed, there is an increasing requirement for networking technology in general to accommodate and expand to satisfy the seemingly insatiable appetite for bandwidth resulting from more users and the deployment of new bandwidth-intensive applications.
The Internet is flourishing. Traffic is increasing at a dramatic rate as the Internet becomes a vital part of the communications infrastructure used today. The Internet has become the cornerstone of the Information Age, and Internet technologies are used extensively for many types of new, large-scale enterprise and private networks.
The types of traffic that flow on these networks now include dizzying arrays of voice, data, and video applications, each having their own particular set of services and resource demands. There is significant momentum in the convergence of the Internet and Public Switched Telephone Network (PSTN) that is introducing many new design, deployment, and management challenges. The number of users and devices that support and access the Internet is exploding. Also, many new and important networking technologies are being introduced and integrated throughout the Internet, from the core backbones all the way down to the edges and home access devices in "the last mile." Fiber optics and optical-based communications are also being introduced to work with packet-based networks with unprecedented urgency and resolve. The core of the Internet is being upgraded with high-speed glass fiber links that connect optical cross-connects (OXCs), wave division multiplexers (WDMs), and soon, photonic switches (PXCs).
A hallmark of the Internet has always been that packet traffic has been delivered on a "best-effort" basis using the TCP/IP protocol suite. Service level requirements are usually met by over-provisioning more and faster links and connecting more numerous and sophisticated routers and switches while assuring that network utilization does not get too high. Conventional routing protocols such as BGP, OSPF, and Intermediate System-Intermediate System (IS-IS) orchestrate routes for the routers placed throughout the network to facilitate the delivery of data packets from source to destination. The routers use these various routing protocol algorithms and signaling techniques to efficiently direct the traffic flow within the Internet.
The new and increased demands for more and different services with varying requirements are quickly forcing network and Internet service providers (NSPs and ISPs, respectively), and other specialized internetworking service providers, to realize that new technologies, in addition to more and faster equipment, are what is needed to keep pace.
Perhaps in the not-too-distant future, when a frustrated network user is pounding the keyboard and asking why the response time is so slow, a network administrator can update the standard reply with a snappier retort such as: "It's your quality of service, stupid."
MPLS is a new network technology that promises to offer QoS and many other new and timely applications to current best-effort, IP-based internetworking. MPLS is designed to operate with the current set of Internet routers and switching equipment, as well as become an important enabling technology for the next generation of Internet devices such as IP terabit (and soon, petabit!) routers and OCX switches. MPLS is also a vital solution that will allow current Layer 2 transport technologies such as ATM, FR, and Ethernet to seamlessly interoperate and coexist with IP-based networks. By allowing new disruptive technologies and current soon-to-be legacy implementations to coexist, MPLS provides network planners flexibility in migration and replacement tactics for their desired networking strategies. MPLS is often spoken of as the "glue" for next-generation networks because of this fact.
MPLS is a hybrid of a traditional network's Layer 3 routing protocols and Layer 2 switching technologies. The concepts are not new, but the way they are combined in MPLS may offer a fresh spin on evolving the Internet routing architecture. MPLS separates the control functions from the data forwarding functions in a manner that allows for new, flexible, and operational efficiencies. MPLS introduces a single forwarding model that is based on label swapping. It offers a richer model for routing packets based on additional and configurable input parameters. It also allows a stack of labels within a packet that can be used in a hierarchical fashion for several innovative, new applications. MPLS can coexist with many existing Layer 2 and Layer 3 protocols and provide scalability for how these protocols are used in today's networks.
MPLS allows for the rapid deployment of multiservice applications, which will allow service providers and network users to realize revenues in new ways. MPLS may be the enabling technology for the next information "gold rush."
MPLS is being developed within a working group of the IETF as an open and standardized way of doing label switching. The MPLS Working Group (WG) is also responsible for defining how to implement label-switched paths over various link level technologies. MPLS includes procedures and protocols for the distribution of labels between the network devices that do label manipulation and swapping.
MPLS was derived mainly from several proprietary efforts that were being done in the early and mid-1990s to marry switching and routing. This allowed MPLS's direction to be expanded into several new areas. Some view MPLS as an attempt to take the best features of the ATM technology and add them to IP. These include the connection-oriented nature of setting up paths and the ability to offer various levels of QoS to packet traffic. MPLS will also allow the interoperability of IP with ATM and FR in an integrated fashion to create a single manageable network infrastructure. Finally, the MPLS label framework is being expanded as a more general model for routing in optical switching networks and additional physical layer technologies, such as time division multiplexing (TDM) and the spatial switching of incoming fiber to outgoing fiber.
To better introduce MPLS, the "five W's and H"a long-standing journalistic device that asks: who, what when, where, why, and howis effective in presenting and summarizing the basics. Because MPLS is integrated as an Internet technology, it is also important to understand the basics of the Internet, its key models, and other closely related topics.
What Is MPLS?
MPLS is an exciting new hybrid technology that integrates the best of the current approaches for delivering packets from their source to their destination across an internetwork1. A formal definition is presented in Figure 112:
Figure 11 A formal definition of MPLS.
By using existing and newly enhanced Internet routing and control protocols, MPLS provides virtual connection-oriented switching across Internet routes by supporting labels and the label swapping paradigm. MPLS includes the implementation of label-switched paths (LSPs) over the most popular link level technologies. MPLS also provides the necessary procedures and protocols for the distribution of labels between MPLS-enabled routers and switches.
MPLS work is being done under the auspices of the MPLS WG within the IETF.3 MPLS is still a very new development; it just entered the Internet standards track in the RFC format at the beginning of 2001. (An RFC is the specification format that is used to place a protocol on the standards track and allow it to progress from proposed to draft to full Internet standard.) Current MPLS RFCs are described in Chapter 3, "MPLS Documentation and Resources." For a full discussion of Internet standards, see Appendix B, "MPLS-Related RFC Index."
Because they are being actively worked on, many MPLS specifications are still largely in Internet draft (ID) format. As such, the exact specifics of the protocol and other ancillary issues are being regularly discussed, revised, and improved. IDs can advance to RFCs. Note that all IDs carry with them the caveat that they are "draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or cite them other than as "works in progress"." The current MPLS IDs as of this writing are also described in Chapter 3, "MPLS Documentation and Resources."
The MPLS WG arose from a consensus at the Tag Switching Birds of a Feather (BOF)4 meeting held at the 37th IETF meeting in December 1996 in San Jose, California.5 The first official meeting of the MPLS WG was held the following April at the 38th IETF meeting in Memphis, Tennessee, where the acronym MPLS first came into common use. The MPLS WG has been working ever since in continuing its efforts to advance the MPLS specifications on the standards track.
The original MPLS WG charter statement is shown in Figure 12. Now that the MPLS architecture and signaling protocols are on the standardization track, the MPLS WG has undertaken additional tasks. These include investigating multicast issues, specifying extensions for authentication, defining and completing SNMP MIB definitions, specifying fault tolerance and recovery mechanisms, and documenting additional encapsulation methods for the latest lower layer technologies.
Figure 12 The original MPLS WG charter.B
Using MPLS for switching TDM slots, optical lambdas (wavelengths), and spatial switching (between incoming and outgoing fibers) are some of the latest developments. These efforts are collectively called Generalized MPLS, or GMPLS. GMPLS is discussed in Chapter 10, "Future MPLS Developments and Directions."
The MPLS WG also originally created a list of eight high-level requirements of "should's" and "must's" to further define the "what" of MPLS:
MPLS should work with most data link technologies.
MPLS should be compatible with most network layer routing protocols and other associated Internet technologies.
MPLS must operate independently from any routing protocols.
MPLS should support a wide range of forwarding granularities for any given label.
MPLS should support operation, administration, and maintenance (OAM).
MPLS must contain loop prevention or detection.
MPLS must operate in a hierarchical network.
MPLS should be very scalable.
These eight requirements are the cornerstones for the major development efforts that have ensued since they were defined. The requirements can be traced through the architecture and the additional supporting specifications.
In conjunction with these requirements, the MPLS WG also set forth a set of eight key objectives. These important MPLS aims are:
The specification of standard protocols to maintain and distribute label binding information to support unicast destination-based routing, where the forwarding is done by swapping labels. (Unicast routing specifies exactly one interface; destination-based routing refers to routing based on the final destination of the packet.)
The specification of standard protocols to maintain and distribute label binding information to support multicast destination-based routing, where the forwarding is done by swapping labels. (Multicast routing specifies more than one interface. The work of integrating multicast techniques and MPLS is still in the discussion phase.)
The specification of standard protocols to maintain and distribute label binding information to support a hierarchy of routing knowledge, where the forwarding is done by swapping labels. (A routing hierarchy refers to knowing the topology within an autonomous system (AS), and also between ASs.)
The specification of standard protocols to maintain and distribute label binding information for supporting explicit paths based on label swapping. These paths may be different from the ones calculated by conventional IP routing, which are based on destination-based forwarding. (Explicit paths are the key to the TE applications that are part of MPLS's prime advantage over conventional routing.)
The specification of standardized procedures for carrying label information over different Layer 2 link level technologies. (MPLS link layer encapsulation makes the technology flexible and scalable.)
The specification of a standard way to interoperate with ATM at the user and control planes (an excellent way to have ATM and IP interoperate, and also a migration path to TCP/IP-only networks).
The specification of support for QoS technologies (such as the Resource Reservation Protocol, or RSVP). (QoS is also one of the most important MPLS applications. MPLS QoS will be a important enabler for the derivation of revenues from next-generation networks.)
The specification of standard protocols to allow hosts to utilize MPLS. (Much like ATM, this objective was never met, as MPLS was implemented from the core outward. Host-based MPLS appears not to be feasible due to scalability and performance issues.)
As can be seen, most of these MPLS objectives serve as the basis for creating and designing the major MPLS technologies and protocols discussed in Chapter 4, "Core MPLS Technologies and Protocols," and Chapter 5, "MPLS Signaling and Label Distribution."