This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: 6.6 Multiprotocol Label Switching 415 to improve this state of affairs so as to offer some quality-of-service (QoS) assurance to the users of the network. Within IP, a mechanism called Diff-Serv (differentiated services) has been proposed. In Diff-Serv, packets are grouped into different classes, with the class type indicated in the IP header. The class type specifies how packets are treated within each router. Packets marked as expedited forwarding (EF) are handled in a separate queue and routed through as quickly as possible. Several additional priority levels of assured forwarding (AF) are also specified. An AF has two attributes: xy . The attribute x typically indicates the queue to which the packet is held in the router prior to switching. The attribute y indicates the drop preference for the packets. Packets with y = 3 have a higher likelihood of being dropped, compared to packets with y = 1 . While Diff-Serv attempts to tackle the QoS issue, it does not provide any end-to- end method to guarantee QoS. For example, we cannot determine a priori if sufficient bandwidth is available in the network to handle a new traffic stream with real-time delay requirements. This is one of the benefits of multiprotocol label switching, which we will study next. 6.6 Multiprotocol Label Switching Multiprotocol label switching (MPLS) is a connection-oriented technology to trans- port IP packets. It has a wide variety of applications. Today MPLS works with other packet-switched networks, providing the same benefits as it does for IP. However, for simplicity, we will focus our discussion on how it works with IP. MPLS can be thought of as a layer sandwiched between the IP layer and the data link layer. MPLS provides a label-switched path (LSP) between nodes in the network. A router implementing MPLS is called a label-switched router (LSR). Each packet now carries a label that is associated with a label-switched path. Each LSR maintains a label-forwarding table, which specifies the outgoing link and outgoing label for each incoming label. When an LSR receives a packet, it extracts the label, uses it to index into the forwarding table, replaces the incoming label with the outgoing label, and forwards the packet on to the link specified in the forwarding table. This very simple MPLS paradigm has several applications in an IP network. 1. Separation of control and data planes: One of the fundamental design philoso- phies in MPLS is that the label-switching and packet-forwarding process at each router is completely decoupled from how LSPs are set up and taken down in the network. We can think of the latter as a network control function, which involves first deciding what LSPs to set up or take down and then actually set- ting them up and taking them down. This simple separation allows us to build 416 Client Layers of the Optical Layer optimized hardware for packet forwarding, independent of the network control mechanisms, and allows for LSPs to be set up and taken down based on different...
View Full Document
This note was uploaded on 01/15/2011 for the course ECE 6543 taught by Professor Boussert during the Spring '09 term at Georgia Tech.
- Spring '09