Multiprotocol Label Switching

Multiprotocol Label Switching (MPLS ) enables the connection-oriented transmission of data packets in a connectionless network along a previously established ( " signaled " ) path. This mediation process is mainly used by operators of large transport networks, voice and data services based on IP to offer (large Internet provider).

3.1 Structure of the MPLS paths
3.2 the passage of IP packets through an MPLS network
Penultimate Hop Popping 3.3

5.1 MPLS label stack entry
5.2 MPLS label stack
5.3 Embedding of the MPLS label stack

Basic idea

MPLS was introduced in order to take advantage of various benefits of connection-oriented switching in otherwise connectionless networks can. For this purpose, on the one hand part of the enabling faster processing of a packet through a simplified addressing using so-called labels; this advantage over traditional longest prefix match forwarding but has since been through advanced technologies ( ASICs) perspective. On the other hand, MPLS allows the network operator with arbitrary paths in its network, which is not traditional routing protocols such as Open Shortest Path First ( OSPF) or IS -IS possible.

Connection-oriented and connectionless transmission

Can data be sent from a terminal spontaneously to a receiver, and know each intermediate node (usually a router) on its own, as he has to forward the data, it is called connectionless data transfer. Must before sending data until a path can be signaled by the network to the receiver by a terminal, it is called connection-oriented data transfer. In this case, the network nodes are provided (usually switches) with the necessary connection information in order to forward the data sent correctly ( label switching ).

A connection-oriented network has a deterministic and controllable behavior. Resources in the switching systems can be reserved during the signaling phase. In contrast, a connectionless network to a stochastic and rather random behavior. In a connectionless network data can arrive in the switching system, which is why the presence of required resources for the transport of certain data can not be guaranteed in all cases at any time and in any quantity.

Historical Development

Even in the mid -1990s prevailed in large-scale communication networks ( WANs), the proportion of voice communication (telephony) significantly the amount of data communication. Due to the above-described differences between connection-oriented (→ telephone calls) and connectionless (→ data packets on the Internet ) operated transmission telecommunications companies separate networks for data and voice transmission, which caused considerable costs. A network wide quality of service (QoS, Quality of Service ) can not exist. Existing voice networks offered this service qualities namely for voice services, but the required bandwidth for data transmission were not available or extremely expensive.

The introduction of ATM solved this problem in many areas. Voice and data could now be transmitted over a common infrastructure. However, the ATM transport network did not provide IP routing functionality for IP-based data transmission (Internet) available. This happened still in routers.

Routing systems through the use of received ATM, however, the ability to use much higher data transfer rates. The signaling of connecting paths is left to the ATM network, while the IP router transmitting connectionless, so their stochastic IP data packets. A network wide quality of service in order to integrate voice and data by using the high-bandwidth, however does not exist. Thereby created so-called overlay architecture in which the IP layer, the underlying ATM transport layer used, but both still operate independently. Examples of this approach are overlay IP over ATM [RFC 2225 ], and Multiprotocol over ATM ( MPoA ).

The available router systems achieved by the newly available high-bandwidth limits of their capacity. In addition, the disassembly and reassembly of IP packets presented ( up to 1536 bytes or more ) into ATM cells ( 53 bytes ), a difficult to overcome limit for speeds over 622 Mbit / S. In Due to the high number of point -to-point connections between the routers, particularly when fully meshed networks, the use of traditional Topologie-/Routing-Protokolle ( IGPs ) such as OSPF, RIP or IS -IS leads to considerable additional signaling traffic ( "n -square- problem ": there are intermediate points in complete meshing edges). This collapse router or develop into permanent bottlenecks in the network. The transmission of different services (voice, data, video) over a uniform and simplified platform does not exist.

MPLS provides solutions to the problem above points.

Basic idea of MPLS

MPLS provides the ability to relieve congested routing systems and thus better utilize the available bandwidth of the long-distance transmission lines since the late 1990s.

The idea is no longer to forward data packets from one router to the next router ( hop-by -hop ), and in each router to meet anew the decision for the best way (complete IP lookup in the so-called Forwarding Table) but to send them to an input point ( ingress router) on a vorsignalisierten data path and to only use again at the starting point ( egress router) the conventional hop-by -hop forwarding of IP. Ideally, the ingress and egress routers are located at the borders of a network. This approach relieves much of the router greatly: In all MPLS -enabled intermediate stations, so-called label switched router (LSR ) is only evaluated the MPLS packets upstream label. This already occurs directly above the data link layer (Layer 2 ) and it is very easy in the appropriate hardware at high speed - in contrast, require forwarding decisions with traditional IP routing the considerably more elaborate longest prefix match.

MPLS provides a connection-oriented traffic as ATM for data packets. The paths are set up once before the packet forwarding ( signaled) and are available from then available. In addition, resources on the routers reserved by means of additional protocols or protocol extensions, such as CR -LDP or RSVP -TE or the routing be influenced. This allows, to a certain extent, to realize QoS for the combined transmission of voice, data and video across the network.

Nevertheless, MPLS can also book with RSVP no bandwidth as it enables ATM. It is possible to approximate a certain deterministic traffic behavior, but IP-Routing/Forwarding is stochastic in his behavior, even with the use of MPLS.

The initial speed advantage of MPLS in data forwarding is now no longer relevant since modern routing systems have consistently implemented the IP forwarding in hardware.

Operation

The use of MPLS into IP networks requires a functioning logical and physical IP-based network infrastructure (MPLS -enabled router). MPLS operates here primarily within the boundaries of a so-called Autonomous System (AS). In addition, the use of an Interior Gateway Protocol ( IGP) such as OSPF or IS -IS, meaningful. Theoretically possible, but not very practical, would also use static routes in combination with IBGP.

Structure of the MPLS paths

After making sure that the routers of an autonomous system (AS) all "see" each other can (which provide such as OSPF or IS -IS safe), now the MPLS paths (paths) between the routers are connected. These paths are called Label Switched Path ( LSP). The initial node of a LSP is referred to as the ingress router, the end point of the egress router. Typically these start and end nodes are located at the input and output points of a AS ( AS Boundary Router).

The switching of the LSPs can be done completely manually, semi-automatically or fully automatically. The manual option requires the configuration of each router, which is an LSP traverses. Wherein autonomous systems in the order of several dozen routers, this method is inefficient. The semi-automatic version only requires the manual configuration of parts of the LSPs, so for example the way over the first three routers. The rest of the pathfinding for the LSP is left to the IGP. The fully automatic version relies in determining the path for an LSP entirely on the IGP. Thus we obtained no advantage in terms of path optimization. However, the data forwarding in the routers is now implemented on Layer 2 (label -swapping, so replacing / changing labels ) rather than Layer 3

Passing IP packets through an MPLS network

When an IP packet enters an MPLS network (see below) it is provided at the ingress router with an additional MPLS header. Considering the ISO layer information ( refer to the ISO / OSI reference model) of a data packet, then the header between the layer 3 information (network layer header ), and the layer - 2 information ( link layer header) inserted. This process of insertion is called push operation. If the label of a LSPs removed by a router, this is called a pop operation. Replacing the labels by a router on the path of LSPs called swap operation.

Penultimate Hop Popping

Penultimate Hop Popping (PHP ) describes the situation in which a MPLS label (when stacked LSPs, the outer label) is removed already in the penultimate routers of an LSP. This so-called PHP router knows, due to the IGP, the path to the egress router and forwards the data packet to that in the normal way on. This saves the egress router, the POP operation, it needs only the extracted packet forwarding based on the routing information.

MPLS today and in the future

The advantage of MPLS is what will happen if additional services are based on MPLS technology, are used. Such - now largely standardized - services are at the present time:

Traffic engineering, targeted control of the paths along which the data traffic of a network. This application allows, for example, a network operator to offer its customers targeted particularly broadband and low-latency data paths. To provide resources for optimized paths through a network, for example, RSVP -TE are used.
Layer 2 VPN: Virtual Private Networks (VPNs) to the OSI layer 2 point- to-point connections. These allow ATM connections to interconnect ( VPI / VCI), Ethernet VLANs or Frame Relay paths ( VC = Virtual Circuits ) of different networks directly over an IP MPLS network. It is the customer handed over a connection on layer 2 of the OSI model at the transfer point. A concrete example would be an Internet service provider that offers DSL access in Germany, but no own nationwide, infrastructure has to connect its ATM-based DSLAMs (Digital Subscriber Line Access Multiplexer) to a centralized BRAS. For this purpose, he uses a different transport providers with IP MPLS infrastructure, the transparent leads the ATM-VPIs/VCIs the central BRAS location. The DSLAMs and BRAS system receive from the transport provider an ATM interface provided, although its infrastructure is purely IP-based. This is also called pseudo -wire emulation ( PWE3 circuits). So lines / paths are emulated. The Ingress-/Egress-Router is referred to in this case as Label Edge Router ( LER ), the routers on the path of LSPs as a Label Switch Router ( LSR). The LSPs are signaled automatically in practice with a separate protocol (such as LDP or L2TPv3 ) between the LERs. But it is also possible to configure the LSPs manually. A Layer 2 VPN is represent as a virtual ATM-/Frame-Relay-/Ethernet-Switch with point-to -point connections
Virtual Private LAN Service ( VPLS ): This is a variant of the Layer 2 VPNs with focus on Ethernet-based infrastructures, ie point-to- multipoint connections, which transmits the broadcast behavior of Ethernet bill. As signaling protocols mainly LDP, BGP but also is used. At the transfer point the customer gets asked a gebridgten LAN port. A VPLS instance represents represent as a virtual LAN switch
Layer 3 VPNs: Virtual Private Networks (VPNs ) on the OSI layer 3 These make it possible to map complete routed network infrastructures of customers transparently over a provider MPLS transport network. It is the customer handed over at the delivery point, a port on Layer 3 of the OSI model, ie a routed connection with a static route or an IGP. More details can be found for example in RFC 4364th The LSPs are signaled in practice by means of LDP. A Layer 3 VPN is as a virtual IP router is ( but what is not to be confused with the proprietary "virtual router " concepts of some manufacturers ).
G- MPLS ( Generalized MPLS) expanded the scope of MPLS to optical transmission infrastructure. This approach is the automatic signaling optical paths include ( so for example, individual lambdas a WDM / DWDM interface, SDH paths or a complete interface ) when a LSPs with. Thus, the signaling of the topology expands its sphere away from the IP transport layer within an AS towards the underlying infrastructure - transport layer. Standardization approaches for the architecture, the functional model and requirements of this can be found under the search terms ASON / ASTN (Automatic Switched Optical Network / Automatic Switched Transport Network).

Basically, the MPLS technology leads the independent packet switching ( connectionless ) back to circuit switching by LSPs (connection oriented ). Thus, some advantages of IP-based communications from any to any ( any-to -any ), restricted with all of their flexibility and good scalability, by the strengths of connection-oriented communication ( complexity, n -square problem, etc.).

Structure of MPLS packets

There are basically two different ways to label a package with MPLS. A, for example, IP, provides for a so-called MPLS shim header is inserted between the layer two header and the layer three header. This header is, however, often dubbed the MPLS label stack (Entry). For connection-oriented networks, however, such as ATM or Frame Relay, the label can be inserted in the Layer 2 header; then there exists no separate MPLS label stack entry.

MPLS Label Stack Entry

The MPLS label stack entry is not a header in the proper sense; Shim the word expresses how short he is. It has a length of 4 bytes (32 bits ), thus generates little overhead and can moreover be processed very quickly. The core of the MPLS label stack Entrys is the MPLS label. The label provides in particular about which path ( LSP Label Switched Path ), the packet should be routed through the MPLS network.

With the 32 bits of the MPLS label stack Entrys four additional information is conveyed:

Labels ( MPLS label, 20 -bit): characteristic information of an LSP (comparable to a telephone number). It is important to understand that this label only has a local scope, ie is only used between two routers on the path of LSPs, not on the entire path from ingress to egress or PHP router.
TC ( Traffic Class; 3 bits): Are used for transmission of Differentiated Services - Information used.
S ( Bottom of Stack, 1 bit): Specifies whether it is at the LSP is a nested LSP, ie whether another LSP is conveyed in the CSP. This specifies the flag to determine whether there are more MPLS labels, or whether this MPLS label stack entry is the last label of the label stack.
TTL ( Time to Live, 8 bits): Defines how many MPLS router must go through the package yet (Limit: 255 router)

MPLS label stack

Typically each packet is assigned a label exactly. However, if you want to nest multiple LSPs into one another, then you can assign more than one label to a MPLS packet also. These are then summarized in the so-called label stack:

Clearly visible here is the use of bottom of stack flag. The evaluation is performed from left to right, according to the " Bottom of Stack" directly follows the Layer 3 header.

Embedding of the MPLS label stack

Depending on whether a nested or a simple CSP is present, an MPLS label stack, consisting of one or many MPLS label stack Entrys, inserted.

Label Distribution Protocol Asynchronous Transfer Mode Network packet Bandwidth (signal processing) Multiprotocol Encapsulation over ATM Teletraffic engineering Resource Reservation Protocol Flag word

254573