Lecture - 24 Multiple Label Switching(MPLS)

New Section

In this section, the speaker discusses the birth of the MPLS paradigm and its advantages over the IP protocol in terms of routing, scalability, and flexibility. However, there were also disadvantages such as performance bottlenecks and difficulties in traffic engineering.

MPLS Paradigm

• The MPLS paradigm was adopted by service providers to address the challenges of traffic engineering.

• It involved having ATM or frame relay based backbone networks with IP routers at the edges.

• Traffic engineering issues were resolved by exploiting the virtual circuit routing capabilities of ATM networks.

Problems with IP over ATM

• IP over ATM led to problems such as segmentation and reassembly overheads due to chopping IP packets into 53-byte ATM cells.

• Excessive network management complexity arose from using separate routing protocols for IP and ATM layers.

• A complete or partial mesh of order n square virtual circuits was required, leading to overheads in routing updates.

Multi-Protocol Label Switching (MPLS)

• To address these problems, it was proposed to have ATM switches participate in IP routing and switch IP packets themselves.

• Forwarding would be based on a label similar to the virtual circuit identifier in ATM cells.

• MPLS combined features from precursors like IP switching by Epsilon, Cisco's tack switching architectures, IBM's Errise architecture, and Toshiba's cell switch router CSR.

New Section

This section focuses on key features of the MPLS protocol or architecture. It explains how paths are established between endpoints using standard IP routing protocols or traffic engineering extensions. Packets are partitioned into forwarding equivalence classes (FEC) at the egress point.

Key Features of MPLS

• Paths between endpoints can be established using standard IP routing protocols or traffic engineering extensions.

• Quality of service attributes can be associated with the established path.

• Packets are partitioned into forwarding equivalence classes (FEC) at the egress of the network.

• The granularity of FEC can vary from a destination IP address prefix to an application flow.

New Section

In this section, the speaker explains how virtual circuit paths, known as label switch paths in MPLS, correspond to forwarding equivalence classes (FEC). The choice between setting up separate label switch paths for individual application flows or aggregating flows towards a destination IP address is discussed.

Correspondence between FEC and Label Switch Path

• There is a correspondence between an FEC and a label switch path in MPLS.

• The choice can be made to set up separate label switch paths for individual application flows or aggregate them towards a destination IP address.

New Section

This section introduces the concept of forwarding equivalence classes (FEC) and labels in MPLS networks. It explains how packets are classified into FECs and associated with labels.

Forwarding Equivalence Classes and Labels

• IP packets at the ingress of the network are classified into forwarding equivalence classes (FEC).

• Each FEC is associated with a label, which maps it to a label switch path.

• In MPLS networks, packets are tagged with a label at the ingress.

• Labels are used in the core of the network to determine the next hop and new label for packet forwarding.

• MPLS follows a label swapping paradigm similar to ATM switches, where labels are swapped instead of IP addresses.

• VC swapping is used in MPLS to address the problem of limited virtual circuit identifiers (VCIs).

• Label switch routers in an MPLS network handle IP packets directly, eliminating fragmentation and reassembly overheads.

• Label switch routers have both control and data components, with the data plane acting on labels for packet forwarding.

• Edge routers participate in both IP forwarding and label switching, classifying packets into FECs and attaching labels.

New Section

This section discusses how MPLS provides an integrated solution for traffic engineering challenges in IP networks. It highlights the benefits of using label switch routers over ATM switches as backbone networks.

Integrated Traffic Engineering Solution

• Label switch routers provide an integrated paradigm for addressing traffic engineering challenges in IP networks.

• Label switch routers can handle both IP routing protocols and label switching.

• Unlike ATM-based switches, label switch routers work directly on IP packets without fragmentation or reassembly overheads.

• The complexity of dealing with separate routing paradigms (ATM-based virtual circuit routing and IP routing) is eliminated.

• MPLS offers a uniform solution for traffic engineering challenges in IP networks.

• Label switch routers have control and data components, with the data plane acting on labels for packet forwarding.

• The label information base in label switch routers contains entries for efficient label lookup and forwarding decisions.

New Section

This section explains the components of a label switch router and how they handle labels for packet forwarding.

Components of a Label Switch Router

• A label switch router consists of a control component and a data component (data plane).

• The data plane of a label switch router operates on labels rather than IP addresses.

• Labels are looked up in the label information base to determine the next hop and the next label to be swapped.

• The size of labels in MPLS is typically smaller than 32-bit IP addresses, reducing the number of entries required in the label information base.

• Index lookups based on labels are efficient, allowing for fast packet forwarding decisions.

• Edge routers participate in both IP forwarding and label switching, classifying packets into FECs and attaching labels.

Please note that these summaries are based solely on the provided transcript.

New Section

This section discusses the process of informing other label switch routers about the bindings created and utilizing them to construct the forwarding table used in label switching.

Label Switching Process

• The label switch router creates bindings between Forwarding Equivalence Classes (FEC) and next hops.

• These bindings are communicated through a label distribution protocol.

• A separate binding is created from labels to next hops.

• Two approaches for informing other routers about the bindings: piggybacking on existing routing protocols or using a separate routing protocol.

• Piggybacking on existing protocols ensures consistency with routing changes but requires extending those protocols.

• Using a separate protocol allows for a clean design but may result in inconsistency between label and routing information bases.

New Section

This section explores the pros and cons of piggybacking on existing routing protocols versus using a separate label distribution protocol.

Piggybacking vs Separate Protocol

• Piggybacking on existing routing protocols ensures consistency in forwarding information base with any topology changes.

• Extending existing protocols can be challenging due to modifications required in multiple messages.

• Using a separate label distribution protocol allows for a clean initial design without relying on legacy routing protocols.

• However, it may lead to inconsistency between the label information base and the routing information base.

New Section

This section discusses MPLS's choice of using a separate label distribution protocol called LDP (Label Distribution Protocol).

LDP as Label Distribution Protocol

• MPLS initially defined LDP as a separate label distribution protocol.

• LDP provides mechanisms for distributing bindings between FECs and labels.

• Later, there was debate regarding whether to use an extension of RSVP or CR-LDP for traffic engineering and quality of service issues.

New Section

This section explains the process of distributing label bindings to all nodes participating in label switching.

Label Distribution Protocol

• At the label edge router, packets are classified into FECs and assigned labels.

• The label distribution protocol is responsible for distributing these label bindings to all participating nodes.

• Network layer routing provides FEC-to-next-hop mapping.

• An FEC can be associated with a destination IP address prefix.

• The network layer routing determines the path through standard routing mechanisms, creating a label-switched path for the associated IP address prefix.

New Section

This section discusses two approaches for creating bindings between FECs and labels: data-driven and control-driven.

Data-driven Approach

• In the data-driven approach, when the first packet of a flow belonging to an IP address prefix arrives at the edge router, it is classified into an FEC and assigned a label.

• Bindings between FECs and labels are created only when packets of that flow arrive for the first time.

The transcript does not provide information on control-driven approach or further details on this topic.

New Section

In this section, the speaker discusses the control-driven approach taken by the Internet Engineering Task Force (IETF) for MPLS standardization.

Control-Driven Approach

• The IETF favored a control-driven approach for MPLS standardization.

• In this approach, Forwarding Equivalence Classes (FECs) are associated with IP address prefixes.

• By utilizing information from the network layer routing, label switch paths corresponding to IP address prefixes can be created in advance.

• Label distribution protocol is used to distribute bindings between FECs and labels to label switch routers.

• This approach does not rely on whether packets bound to a particular IP address prefix appear at the edge of the network.

New Section

This section explains the structure of MPLS labels and their significance in forwarding packets.

Structure of MPLS Labels

• MPLS labels consist of a 4-byte shim header.

• The 32-bit label includes a 20-bit identifier or index and a 3-bit exp field for experimental purposes.

• There is also a field called TTL (Time-to-Live), which indicates if there is a label stack following it.

• The TTL field is important for loop prevention in MPLS networks.

New Section

This section discusses how TTL fields are used in MPLS networks and their role in preventing loops.

Usage of TTL Fields

• In an MPLS network, routers in the core only forward based on MPLS labels, not IP headers.

• To ensure proper forwarding, the value of the TTL field from the IP header is copied into the MPLS label at ingress.

• When reaching egress, the TTL field is copied back into the IP header before further processing by IP routers.

• If the TTL field reaches zero in either the MPLS label or IP header, the packet is dropped to prevent loops.

New Section

This section explains various components of MPLS, including Forwarding Equivalence Class (FEC), Next Hop Label Forwarding Entry (NHLFE), and Incoming Label Map.

Components of MPLS

• Forwarding Equivalence Class (FEC) represents a set of classification rules for forwarding.

• The granularity of forwarding determines network scalability.

• At the core level, MPLS can use IP prefixes for destination-based forwarding.

• Next Hop Label Forwarding Entry (NHLFE) maps FECs to next hops and is obtained from network layer routing.

• Incoming Label Map provides bindings between labels and next hops.

• MPLS has a data plane for labeled packet travel and a control plane for exchanging label binding information.

New Section

This section discusses control protocols used in MPLS networks and the importance of forwarding granularity.

Control Protocols and Forwarding Granularity

• Control protocols like RSVP-TE and CR-LDP are used for exchanging label binding information in MPLS networks.

• Forwarding Equivalence Class (FEC) determines the set of classification rules for forwarding packets.

• The choice of forwarding granularity affects network scalability.

• At the core level, IP prefixes can be used for destination-based forwarding.

MPLS Label Switching and Scalability

In this section, we discuss the scalability of MPLS-based networks and the use of labels for forwarding.

Scalability of MPLS-based Networks

• Associating a label switch path with each application flow is not scalable due to the large number of flows in the network.

• Alternatively, associating a label switch path with an IP address prefix and using destination-based forwarding is more scalable.

MPLS Label Structure

• The MPLS label is a 32-bit shim header inserted between the IP header and link layer header.

• Routers work only on the MPLS label, not on the IP header.

• The TTL field from the IP header is copied into the MPLS label.

Basic Operations of Label Switch Routers

• Label switch routers perform three basic operations: push, pop, and swap.

• Swap operation determines the next hop based on incoming label and forwards the packet while swapping it with an outgoing label.

• Push and pop operations are required when using a stack of labels.

Bundling LSPs in MPLS Networks

This section explains how LSPs can be bundled or nested in MPLS networks for improved efficiency.

Bundling LSPs

• Similar to ATM networks, multiple virtual circuits can be bundled into one virtual path in MPLS networks.

• In MPLS, bundling can occur at arbitrary levels by nesting smaller LSPs into larger ones.

• Each level of bundling is identified by an outer label followed by an inner label for individual LSPs.

Stack of Labels

• Unlike ATM networks with two levels (virtual circuit or virtual path), MPLS allows arbitrary levels of stacking by increasing the size of the label stack.

• The bottom-of-stack (S) bit indicates the bottom of the stack, and labels can be pushed or popped accordingly.

Egress and Ingress Operations

• At the egress, the outer label is popped, and packets are forwarded based on the inner label.

• At points where LSP nesting occurs, a push operation is performed to create a stack of labels.

Manipulating Labels in MPLS Networks

This section discusses how multiple control planes can manipulate labels on a single packet in MPLS networks.

Label Associated with Forwarding Treatment

• A label can be associated with a label switch path and used to identify the forwarding treatment for a packet.

• For example, it can determine which queue the packet should be queued in for appropriate quality of service scheduling.

Multiple Control Planes

• Multiple control planes can manipulate labels on a single packet.

• This allows different control planes to use labels for various purposes such as forwarding treatment identification.

The transcript provided does not include any timestamps beyond 2258 seconds.

Independent vs Ordered Label Distribution

This section explains the difference between independent and ordered label distribution in MPLS networks.

Independent Label Distribution

• Each Label Switch Router (LSR) independently decides to assign a label to Forwarding Equivalence Classes (FECs) and advertise it to its neighbors.

• This approach allows for fast label distribution and easy creation of specific label switch paths.

• However, inconsistencies in FEC definitions among LSRs can lead to the inability to establish certain label switch paths.

Ordered Label Distribution

• In an ordered approach, label assignment proceeds in an orderly fashion from one end of the Label Switched Path (LSP) to another.

• This ensures consistency in choices made by LSRs but results in slower convergence time compared to independent distribution.

• It is suitable for networks transitioning from traditional networks to MPLS-based networks.

Advantages and Disadvantages of Independent and Ordered Approaches

This section discusses the advantages and disadvantages of both independent and ordered approaches in MPLS networks.

Advantages of Independent Approach

• Fast label distribution mechanism enables quick creation of specific label switch paths.

• Works well when network stability is achieved, consistent FEC definitions are understood by all routers, and there is agreement on network administration mechanisms or provisioning.

Advantages of Ordered Approach

• Ensures no inconsistencies or different choices made by LSRs during label assignment.

• Suitable for networks transitioning from traditional networks to MPLS-based networks.

Disadvantages of Independent Approach

• Inconsistent FEC definitions among LSRs may prevent establishing certain label switch paths.

Disadvantages of Ordered Approach

• Slower convergence time compared to independent approach.

Label Stacking and Aggregation

This section explains how multiple labels can be encoded to form a label stack in MPLS networks.

• Multiple labels can be encoded to create nested Label Switched Paths (LSPs) similar to virtual circuits or IP in IP tunnels.

• Push and pop operations are used to aggregate multiple LSPs into a single LSP tunnel.

• Each Label Switch Router performs push and pop operations based on the label bindings.

Downstream and Upstream Based Label Allocation

This section discusses downstream and upstream based label allocation in MPLS networks.

Downstream Based Label Allocation

• Labels are allocated from the destination towards the source of packet flow.

• The egress router assigns an incoming label, which is then used by the ingress router for forwarding.

Upstream Based Label Allocation

• Labels are allocated in the same direction as the packet flow.

• The ingress router assigns an outgoing label, which is used by the next hop router for forwarding.

Summary of MPLS Functionality

This section provides a summary of MPLS functionality and its application in network edge routers.

• At the edge of an MPLS network, routers have capabilities for both IP forwarding and label switching.

• Packets at the edge are classified into Forwarding Equivalence Classes (FECs).

• FECs can be associated with destination IP-based forwarding or other classification rules.

• Labels are bound to FECs, and label distribution mechanisms ensure proper forwarding within the network.

Label Distribution in MPLS Networks

This section discusses the different approaches and protocols used for distributing label bindings in MPLS networks. It also explains how label switch paths are created and how intermediate nodes forward packets based on labels.

Approaches for Label Distribution

• Different approaches exist for distributing label bindings in MPLS networks.

• Protocols can piggyback on existing routing protocols or use separate label distribution protocols.

Forwarding Based on Labels

• Intermediate nodes (label switch routers) in an MPLS network forward packets based only on labels.

• Labels can be used to identify the next hop, next label, and quality of service attributes.