Tag Switching technology

Review

Status of this memo

This article is a draft of the Internet. The draft Internet is the working document for the Internet Engineering task Force (IETF) and the areas and workgroups it covers. Note that other workgroups can also publish working documents as Internet drafts.

The Internet draft is valid for a maximum of 6 months, a draft document that is likely to be modified, replaced or repealed at any time by other documents. It is not appropriate to use the Internet draft as a reference material or to refer to them in the form of "working in progress".

If you want to know the current state of any of the Internet drafts, Please refer to included in ftp.is.co.za (Africa), nic.nordu.net (Europe), munnari.oz.au (Pacific Rim), Ds.internic.net (United States East Coast) or ftp.isi.edu (United States West Coast), internet-drafts The "lid-abstracts.txt" file in the Shadow directory.

Summary

This paper introduces a technique called tag switching, which is a new method for network layer information packet forwarding. This paper mainly describes the two main components of this tag switching architecture: forwarding and control. Forwarding is implemented through the application of simple label switching techniques, while existing network layer routing protocols are used in conjunction with the mechanism of the binder and allocation tags for control. Tag switching maintains the extended properties of IP and helps to improve the scalability of the IP network. Tag switching does not depend on ATM, and it can be applied directly to ATM switches. This paper also introduces the application and deployment of a series of tag exchanges.

Introduction

The continued development of the Internet increasingly requires Internet service providers (ISPs) to provide higher bandwidth. However, the development of the Internet is not the only driver of higher bandwidth-the need for higher bandwidth, but also the growing popularity of multimedia applications.

Higher bandwidth requirements also require that routers have higher forwarding performance (packets per second) for both multicast and unicast traffic.

The development of the Internet also requires the Internet routing system to improve its extended performance. The ability to maintain a large number of routing information by a single router and the ability to build a routing knowledge layer are essential to support a high-quality, scalable routing system.

We have become aware of the need to improve forwarding performance while increasing routing capabilities to support multicast, allowing more flexibility in controlling traffic routing and providing the ability to build a routing knowledge layer. Moreover, it is becoming increasingly important to have a routing system that supports moderate development to meet new needs.

Tag swapping is an effective solution to these challenges. Tag Exchange combines the flexibility and richness of network layer routing with the simplicity of the label switching forwarding model. The simplicity of the tag switching forwarding mode (tag switching) improves forwarding performance while maintaining a competitive performance price ratio. By linking a large range of forwarding granularity to tags, the same forwarding pattern can be used to support a wide variety of routing functions, such as destination based routing, multicast, routing knowledge layer and flexible routing control. Finally, with the same forwarding mode, a simple forwarding, a large range of forwarding granularity and the ability to develop the routing function combined, there will be to ensure a moderate development to adapt to the emerging needs of the routing system.

The remainder of this document is arranged as follows: Section 2nd introduces the main components of marking Exchange-forwarding and control. The 3rd section describes the forwarding part. The 4th section describes the control components. The 5th section explains how token swapping is used with ATM. The 6th section describes the use of token exchange assistance to provide a range of quality of service. The 7th section briefly describes the possible configuration scenarios. The 8th section concludes.

Mark Swap Parts

The tag exchange consists of two parts: forwarding and control. The forwarding part completes the forwarding of packets by using the tag information (tags) that the packet carries and the token forwarding information maintained by the tag switch. The control part is responsible for maintaining the correct token forwarding information in a set of interconnected tag switches.

Forwarding part

The basic forwarding pattern used for tag switching is based on the idea of label swapping. When a tagged packet is received by a tagged switch, the switch uses this tag as an index in its tag information Base (TIB). Each entry in the TIB includes a stop mark, one or more table entries (outbound tags, outbound interfaces, outbound link-level information). If the switch finds that a inbound tag for an item is the same as the tag carried by the packet, the switch processes each subkey (outbound flag, outbound interface, outbound link level information) for that item, replacing the tag in the packet with an outbound token, Replace the link level information (such as MAC address) in the packet with the outbound link level information. And through the outbound interface, the packet is forwarded out.

From the above introduction of the forwarding part, we can find the following points: first, the forwarding decision is based on a precise matching algorithm, which is a fixed length, rather short tags as the index. It uses a simplified forwarding process relative to the traditional longest matching forwarding for the network layer.

This, in turn, ensures higher forwarding performance (more packets per second). This forwarding process is simple enough to allow direct hardware implementation. The 2nd discovery is that forwarding decisions have nothing to do with the forwarding granularity of tokens. For example, the same forwarding algorithm applies to multicast and unicast-a single broadcast item will only have a single (outbound tag, outbound interface, outbound link-level information) subkey, and a multicast will have one or more (outbound, outbound, Outbound link-level information) subkeys. (For multiple Access links, in this situation, outbound link level information will include a multicast MAC address). This illustrates how the same forwarding pattern uses tag switching to support different routing functions (such as multicast, unicast, and so on).

This simple forwarding process must not be coupled to the control part of the token exchange. The new Routing (Control) feature can be easily deployed without interfering with forwarding mode. This means that when new routing features are added, there is no need to optimize forwarding performance (by modifying hardware or software).