Basic Operation of MPLS TE
The operation of MPLS TE involves link information distribution, path computation, LSP signaling, and traffic selection. This section explains the most important concepts behind each of these four steps. LSRs implement the first two steps, link information distribution and path computation, when they need perform constraint-based routing. MPLS networks that do not use constraint-based routing (or use an offline tool for that purpose) perform only LSP signaling and traffic selection. MPLS TE does not define any new protocols even though it represents a significant change in how MPLS networks can route traffic. Instead, it uses extensions to existing IP protocols.
Link Information Distribution
MPLS TE uses extensions to existing IP link-state routing protocols to distribute topology information. An LSR requires detailed network information to perform constraint-based routing. It needs to know the current state of an extended list of link attributes to take a set of constraints into consideration during path computation for a TE LSP. Link-state protocols (IS-IS and OSPF) provide the flooding capabilities that are required to distribute these attributes. LSRs use this information to build a TE topology database. This database is separate from the regular topology database that LSRs build for hop-by-hop destination based routing.
MPLS TE introduces available bandwidth, an administrative group (flags), and a TE metric as new link attributes. Each link has eight amounts of available bandwidth corresponding to the eight priority levels that TE LSPs can have. The administrative group (flags) acts as a classification mechanism to define link inclusion and exclusion rules. The TE metric is a second link metric for path optimization (similar to the IGP link metric). In addition, LSRs distribute a TE ID that has a similar function to a router ID. The OSPF and IS-IS extensions mirror each other and have the same semantics. Table 2-1 shows the complete list of link attributes. RFC 3784 and RFC 3630 define the IS-IS and OSPF extensions for TE respectively.
Table 2-1. Extended Link Attributes Distributed for TE
|
Link Attribute |
Description |
|
Interface address |
IP address of the interface corresponding to the link |
|
Neighbor address |
IP address of the neighbor's interface corresponding to the link |
|
Maximum link bandwidth |
True link capacity (in the neighbor direction) |
|
Reservable link bandwidth |
Maximum bandwidth that can be reserved (in the neighbor direction) |
|
Unreserved bandwidth |
Available bandwidth at each of the (eight) preemption priority levels (in the neighbor direction) |
|
TE metric |
Link metric for TE (may differ from the IGP metric) |
|
Administrative group |
Administratively value (flags) associated with the link for inclusion/exclusion policies |
MPLS TE can still perform constraint-based routing in the presence of multiple IGP areas or multiple autonomous systems. OSPF and IS-IS use the concept of areas or levels to limit the scope of information flooding. An LSR in a network with multiple areas only builds a partial topology database. The existence of these partial databases has some implications for path computation, as the next section describes. LSRs in an inter-autonomous system TE environment also need to deal with partial network topologies. Fortunately, inter-area TE and inter-autonomous system TE use similar approaches for constraint-based routing in the presence of partial topology information.
Path Computation
LSRs can perform path computation for a TE LSP using the TE topology database. A common approach for performing constraint-based routing on the LSRs is to use an extension of the shortest path first (SPF) algorithm. This extension to the original algorithm generally receives the name of constraint-based, shortest path first (CSPF). The modified algorithm executes the SPF algorithm on the topology that results from the removal of the links that do not meet the TE LSP constraints. The algorithm may use the IGP link metric or the link TE metric to determine the shortest path. CSPF does not guarantee a completely optimal mapping of traffic streams to network resources, but it is considered an adequate approximation. MPLS TE specifications do not require that LSRs perform path computation or attempt to standardize a path computation algorithm.
Figure 2-2 illustrates a simplified version of CSPF on a sample network. In this case, node E wants to compute the shortest path to node H with the following constraints: only links with at least 50 bandwidth units available and an administrative group value of 0xFF. Node E examines the TE topology database and disregards links with insufficient bandwidth or administrative group values other than 0xFF. The dotted lines in the topology represent links that CSPF disregards. Subsequently, node E executes the shortest path algorithm on the reduced topology using the link metric values. In this case, the shortest path is {E, F, B, C, H}. Using this result, node E can initiate the TE LSP signaling.
)
Figure 2-2 Path Computation Using the CSPF Algorithm
Path computation in multi-area or inter-autonomous system environments may involve several partial computations along the TE LSP. When the headend does not have a complete view of the network topology, it can specify the path as a list of predefined boundary LSR (Area Border Router [ABR] in the case of inter-area and Autonomous System Boundary Router [ASBR] in the case of inter--autonomous system). The headend can compute a path to the first boundary LSR (which must be in its topology database and initiate the signalling of the TE LSP signaling can be initiated). When the signaling reaches the boundary LSR, that LSR performs the path computation to the final destination if it is in its topology. If the destination is not in the topology, the signaling should indicate the next exit boundary LSR, and the path computation will take place to that boundary LSR. The process continues until the signaling reaches the destination. This process of completing path computation during TE LSP signaling is called loose routing.
Figure 2-3 shows path computation in a sample network with multiple IGP areas. All LSRs have a partial network topology. The network computes a path between nodes E and H crossing the three IGP areas in the network. Node E selected nodes F and G, which have as the boundary LSRs that the TE LSP will traverse. Node E computes the path to node F and initiates the TE LSP signaling. When node F receives the signaling message, it computes the next segment of the path toward node G. When the signaling arrives at node G, it completes the path computation toward node H in area 2. The next section explains how LSRs signal TE LSPs.
)
Figure 2-3 Multi-Area Path Computation
Signaling of TE LSPs
MPLS TE introduces extensions to RSVP to signal up LSPs. RSVP uses five new objects: LABEL_REQUEST, LABEL, EXPLICIT_ROUTE, RECORD_ROUTE, and SESSION_ATTRIBUTE. RSVP Path messages use a LABEL_REQUEST object to request a label binding at each hop. Resv messages use a LABEL object to perform label distribution. Network nodes perform downstream-on-demand label distribution using these two objects. The EXPLICIT_ROUTE object contains a hop list that defines the explicit routed path that the signaling will follow. The RECORD_ROUTE object collects hop and label information along the signaling path. The SESSION_ATTRIBUTE object lists the attribute requirements of the LSP (priority, protection, and so forth).
RFC 3209 defines these RSVP TE extensions. Table 2-2 summarizes the new RSVP objects and their function.
Table 2-2. New RSVP Objects to Support MPLS TE
|
RSVP Object |
RSVP Message |
Description |
|
LABEL_REQUEST |
Path |
Label request to downstream neighbor |
|
LABEL |
Resv |
MPLS label allocated by downstream neighbor |
|
EXPLICIT_ROUTE |
Path |
Hop list defining the course of the TE LSP |
|
RECORD_ROUTE |
Path, Resv |
Hop/label list recorded during TE LSP setup |
|
SESSION_ATTRIBUTE |
Path |
Requested LSP attributes (priority, protection, affinities) |
Figure 2-4 illustrates the setup of a TE LSP using RSVP. In this scenario, node E signals a TE LSP toward node H. RSVP Path messages flow downstream with a collection of objects, four of which are related to MPLS TE (EXPLICIT_ROUTE, LABEL_REQUEST, SESSION_ATTRIBUTE, and RECORD_ROUTE). Resv messages flow upstream and include two objects related to MPLS TE (LABEL and RECORD_ROUTE). Each node performs admission control and builds the LSP forwarding information base (LFIB) when processing the Resv messages. The structure of the LFIB and the MPLS forwarding algorithm remain the same regardless of the protocols that populated the information (for example, RSVP in the case of MPLS TE).
)
Figure 2-4 TE LSP Setup Using RSVP
Traffic Selection
MPLS TE separates TE LSP creation from the process of selecting the traffic that will use the TE LSP. A headend can signal a TE LSP, but traffic will start flowing through the LSP after the LSR implements a traffic-selection mechanism. Traffic can enter the TE LSP only at the headend. Therefore, the selection of the traffic is a local head-end decision that can use different approaches without the need for standardization. The selection criteria can be static or dynamic. It can also depend on the packet type (for instance, IP or Ethernet) or packet contents (for example, class of service). An MPLS network can make use of several traffic-selection mechanisms depending on the services it offers.