Tag Switching

This chapter describes tag switching, a high-performance, packet-forwarding technology developed by Cisco Systems. Tag switching combines the benefits of routing with the performance of switching.

Note The information in this chapter is applicable to the Catalyst 8540 MSR, Catalyst 8510 MSR, and LightStream 1010 ATM switch router. For detailed configuration information, refer to the ATM Switch Router Software Configuration Guide and the ATM Switch Router Command Reference publication.

This chapter includes the following sections:

Overview

Tag switching, which integrates network layer (Layer 3) routing and data link layer (Layer 2) switching, provides scalable, high-speed switching in the network core. Tag switching technology is based on the concept of label swapping, in which packets or cells are assigned short, fixed-length labels that tell switching nodes how data should be forwarded.

Tag switching provides additional benefits in the areas of functionality, scalability, traffic management, and flexibility for service providers. For example, tag switching offers a flexible and scalable method to provide virtual private network (VPN) services with QoS. The traffic engineering features of tag switching are useful for managing traffic and link utilization in a routed network. Finally, tag switching's ability to integrate ATM and IP technology is of interest to those who want to use an ATM backbone to build a multiservice network.

The Internet Engineering Task Force (IETF) is developing a standard for tag switching based on Cisco's technology. The IETF term for its tag switching standard is Multiprotocol Label Switching (MPLS).

Tag Switching Components

A tag switching network consists of two types of devices, tag edge routers and tag switches (see Figure 10-1).

Figure 10-1: Tag Switching Network

Tag Edge Routers

Tag edge routers are network layer routing devices located at the edges of a tag switching network. Tag edge routers examine packets entering the tag switching network and apply the proper tag, or label, to the packet before forwarding it to the next hop. For packets leaving the tag switching network, tag edge routers perform the reverse function, removing tags from packets. Tag edge routers also perform value-added network layer services such as security, accounting, and QoS classification.

Tag edge routers use standard routing protocols to create routing tables, which identify routes through the network. Based on the routing tables, tag edge routers use the Tag Distribution Protocol (TDP) to apply and distribute tags to other tag edge routers or tag switches.

Tag Switches

At the core of a tag switching network are tag switches, which forward tagged packets or cells based on tags. ATM switches can be used as tag switches, allowing lookup and forwarding capabilities using fast hardware techniques. Because tag switching and ATM Forum-compliant ATM can coexist on the same ATM switch, your ATM switch router can provide both tag switching and ATM services in parallel. Standard routers, equipped with the proper software, can also function as tag switches.

Tag switches receive TDP information from the tag edge routers and build their own forwarding database. Tag switches then switch the packets based on the tags only (VPI/VCI in the case of ATM), without looking at the Layer 3 header.

Tag Distribution Protocol

The Tag Distribution Protocol (TDP) is used by tag switching devices to distribute, request, and release tag binding information for the IP protocol in a tag switching network. TDP does not replace a routing protocol. Instead, it uses information learned from the routing protocol to create tag bindings.

Information Components

Tag switching utilizes three types of information bases for storing and retrieving forwarding information:

Forwarding Information Base (FIB)---a condensed form of routing table containing the destination address, next-hop address, and outgoing interface. Routers make forwarding decisions based on the destination address of a packet plus the information in the FIB.
Tag Information Base (TIB)---serves to bind an incoming tag to one or more of the following:
- Outgoing tag
- Destination address
- Outgoing link-level information

: A TIB can exist for a whole switch, an interface, or a combination of the two.

Tag Forwarding Information Base (TFIB)---uses information from the FIB and TIB to construct the forwarding information, consisting of the incoming interface, destination address, incoming tag, most efficient next hop, and outgoing interface.

How it Works

In conventional Layer 3 forwarding, as a packet traverses the network, each router extracts forwarding information from the Layer 3 header. Header analysis is repeated at each router (hop) through which the packet passes.

In a tag switching network, the Layer 3 header is analyzed just once. It is then mapped into a short fixed-length tag. At each hop, the forwarding decision is made by looking only at the value of the tag. There is no need to reanalyze the Layer 3 header. Because the tag is a fixed-length, unstructured value, lookup is fast and simple.

Figure 10-2 illustrates the operation of a tag switching network.

Figure 10-2: Tag Switching Operation

When a tag edge router at the entry point of a tag switching network receives a packet for forwarding, the following process occurs:

1. Tag edge routers and tag switches use standard routing protocols to identify routes through the network. This routing information is summarized in the FIB.

2. Tag switches use the tables generated by the standard routing protocols to assign and distribute tag information using TDP. Tag switches receive TDP information and build the TFIB that makes use of the tags.

3. When a tag edge router receives a packet for forwarding across the network, it does the following:

(a) Analyzes the network-layer header

(b) Performs applicable network-layer services

(d) Applies a tag and forwards the packet to the next-hop tag switch

4. The tag switch receives the tagged packet and switches the packet based solely on the tag, without reanalyzing the network-layer header.

5. The packet reaches the tag edge router at the egress point of the network, where the tag is stripped off and the packet delivered.

Tag Switching in ATM Environments

Because both tag switching and ATM switching forward traffic based on label swapping, tag switching can readily be applied to ATM switching environments. In addition, ATM switches can use tag switching and still perform ATM Forum standard User-to-Network (UNI) signaling and Private Network-to-Network (PNNI) routing functions.

Advantages

Advantages of implementing tag switching in an ATM network include the following:

Combines the performance and traffic management capabilities of Layer 2 (data link layer) switching with the scalability and flexibility of Layer 3 (network layer) routing. You can direct packet flows across a router-based network along predetermined paths, such as virtual circuits, rather than hop-by-hop as in a typical router-based network. This allows routers to perform advanced traffic management tasks, such as load balancing, in ATM switch routers.
Uses standard routing protocols and TDP to distribute tag values with other tag switches and with tag edge routers. Unlike standard ATM, there is no call setup procedure.
There is no need to use switched virtual circuits (SVCs) for dynamic IP packet flows. This frees CPU processing power for PNNI and UNI flows, which can be reserved for transferring real-time voice and video and other time-sensitive traffic.
Implementing tag switching on an ATM switch does not preclude the ability to support an ATM control plane, such as PNNI, on the same switch. On physical interfaces on the ATM switch router, tag switching and the ATM control plane operate in a ships in the night (SIN) mode and are unaware of each other.
There are no high call setup rates. Standard ATM uses a connection setup procedure to allocate VCIs and program the ATM switching hardware, but tag switching uses standard routing protocols and TDP.
ATM switches appear as routers to adjacent routers. This provides a scalable alternative to the overlay model (ATM switches in the core network and routers on the periphery) and removes the necessity for ATM addressing, routing, and signaling schemes.
ATM switches can participate fully in hierarchical routing protocols and act as peers to tag edge routers. Thus the tag edge routers see far fewer peers than if the edge routers were interconnected via virtual circuits over the ATM network. Therefore, in terms of the number of network devices, the network can scale to much larger sizes.

Limitations

Limitations of tag switching include the following:

Only the Open Shortest Path First (OSPF) routing protocol is supported in the tag switching implementation on the ATM switch router.
Understanding tag switching at the troubleshooting level requires a knowledge of OSPF, the Border Gateway Protocol (BGP), and Cisco Express Forwarding (CEF).

Hardware and Software Requirements and Restrictions

Tag switching has certain hardware requirements on the ATM switch router. For specifics, refer to the ATM Switch Router Software Configuration Guide.

Tag switching on the ATM switch router has the following software restrictions:

OSPF is the only routing protocol currently supported
IP is the only network layer protocol supported.
Hierarchical VP tunnels and tag switching cannot co-exist on a physical interface.
No tag switching virtual circuits (or any other virtual circuits) can be set up on a physical interface with a hierarchical VP tunnel.

Tag Switching Configuration Overview

This section provides a high-level overview of configuring tag switching on ATM switch routers. Following is a summary of the tasks:

Step 1 Configure a loopback interface.

Step 2 Enable tag switching on the ATM interface.

Step 3 Configure OSPF.

Step 4 Configure a VPI range (optional).

Step 5 Configure TDP control channel (optional).

Step 6 Configure tag switching on VP tunnels.

Step 7 Configure VC merge (optional).

Step 8 Configure CoS.

The Loopback Interface

You should configure a loopback interface on every ATM switch router configured for tag switching. The loopback interface, a virtual interface, is always active. The IP address of the loopback interface is used as the TDP identifier for the ATM switch router. If a loopback interface does not exist, the TDP identifier is the highest IP address configured on the ATM switch router. If that IP address is administratively shut down, all TDP sessions through the ATM switch router restart. Therefore, we recommend that you configure a loopback interface.

Configuring the loopback interface requires the following steps:

Step 1 Enter interface configuration mode and assign a number to the loopback interface.

Step 2 Assign an IP address and subnet mask to the loopback interface.

Tag Switching on the ATM Interface

Enabling tag switching on the ATM interface requires the following steps:

Step 1 Select the ATM interface to configure and enter interface configuration mode.

Step 2 Do one of the following:

(a) Enable IP unnumbered on the ATM interface and assign the unnumbered interface to an interface that has an IP address. This is the recommended method. It allows you to conserve IP addresses and reduces the number of TVCs terminating on the switch.

(b) Assign an IP address and subnet mask to the ATM interface.

All parallel interfaces between ATM switch routers should be configured with the same method.

Step 3 Enable tag switching of IPv4 packets on the interface.

The Routing Protocol

OSPF must be enabled on the ATM switch router so that it can create the routing tables used to identify routes through the network. Addresses and associated routing areas are then added to the OSPF process so that it can propagate the addresses to other ATM switch routers.

Configuring OSPF requires the following steps:

Step 1 Enable OSPF and assign it a process number.

Step 2 Define the network prefix, a wildcard subnet mask, and the associated area number on which to run OSPF.

Repeat this step for each additional area you want to add to the OSPF process.

The VPI Range

You might need to change the default tag virtual path identifier (VPI) range on the switch if:

It is an administrative policy to use a VPI value other than 1, the default VPI.
There is a large number of TVCs on an interface.

Configuring a VPI range requires the following steps:

Step 1 Select the interface to configure and enter interface configuration mode.

Step 2 Enter the VPI range.

Note You cannot enter a VPI range on a VP tunnel. On VP tunnels, the VPI is the permanent virtual path (PVP) number of the tunnel.

The TDP Control Channel

You can change the default TDP control channel VPI and VCI if you want to use a nondefault value. The default TDP control channel is on VPI 0 and VCI 32. TDP control channels exchange TDP HELLOs and Protocol Information Elements (PIEs) to establish two-way TDP sessions. Tag virtual channels (TVCs) are created by exchanging PIEs through TDP control channels.

Figure 10-3 shows an example TDP control channel configuration between a source switch and destination switch on ATM interface 0/0/1. Note that the VPI and VCI values match on the source switch and destination switch.

Figure 10-3: Configuring TDP Control Channels

Changing the TDP control channel requires the following steps:

Step 1 Select the interface to configure and enter interface configuration mode.

Step 2 Specify the new VPI and VCI values for the new TDP control channel configuration.

Tag Switching on VP Tunnels

You can configure tag switching on VP tunnels. To do so, you must first configure the VP tunnel on the source and destination switch, then connect the VP tunnel at the intermediate switch.

Figure 10-4 shows an example VP tunnel between a source switch and destination switch.

Figure 10-4: Configuring VP Tunnels

Configuring tag switching on a VP tunnel requires the following steps:

Step 1 Select the interface to configure and enter interface configuration mode.

Step 2 Create a PVP with a VPI value.

Step 3 Select the interface and subinterface you specified in the previous steps.

Step 4 Enable IP unnumbered or assign an IP address and subnet mask to the subinterface, as described in Step 2 in the "Tag Switching on the ATM Interface" section.

Step 5 Enable tag switching of IPv4 packets on the interface.

Repeat these steps on the ATM switch router at the other end of the VP tunnel.

To complete the VP tunnel, you must configure the ATM ports on the intermediate switch to designate where to send packets coming from the source switch and going to the
destination switch.

Figure 10-5 shows an example configuration on an intermediate switch.

Figure 10-5: Connecting the VP Tunnels

Configuring the cross-connect in the intermediate switch requires the following steps:

Step 1 Select one of the interfaces and enter interface configuration mode.

Step 2 Connect the PVP from the source switch to the destination switch by specifying the PVP on this interface, the number of the opposite interface, and the PVP to use on that end.

VC Merge

Virtual circuit (VC) merge allows the switch to aggregate multiple incoming flows with the same destination address into a single outgoing flow. Where VC merge occurs, several incoming tags are mapped to one single outgoing tag. Cells from different VCIs going to the same destination are transmitted to the same outgoing VC using multipoint-to-point connections. This sharing of tags reduces the total number of virtual circuits required for tag switching. Without VC merge, each source-destination prefix pair consumes one tag VC on each interface along the path. VC merge reduces the tag space shortage by sharing tags for different flows with the same destination.

Note There are specific hardware requirements for the VC merge feature. For specifics, refer to the ATM Switch Router Software Configuration Guide.

Figure 10-6 shows an example of VC merge. In Figure 10-6, routers A and B are sending traffic to prefix 171.69.0.0/16 on router C. The ATM switch router in the middle is configured with a single outbound VCI 50 bound to prefix 171.69.0.0/16. Data flows from routers A and B congregate in the ATM switch router and share the same outgoing VC. Cells coming from VCIs 40 and 90 are buffered in the input queues of the ATM switch router until complete AAL5 frames are received. The complete frame is then forwarded to router C on VCI 50.

Figure 10-6: VC Merge Example

VC merge is enabled by default on the ATM switch router. No manual configuration is required for it to work.

Tag Switching CoS

Class of service (CoS) is supported for tag switching on the ATM switch router. For related information on ATM QoS classes, see "Traffic and Resource Management."

Note Tag switching support for CoS has specific hardware requirements. For details, refer to the ATM Switch Router Software Configuration Guide.

With tag switching CoS, tag switching can be dynamically set up to four tag virtual channels (TVCs) with different service categories between a source and destination. TVCs do not share the same QoS classes reserved for ATM Forum VCs (VBR-RT, VBR-NRT, ABR, and UBR). The following four new service classes were created for TVCs: TBR_1 (WRR_1), TBR_2 (WRR_2), TBR_3 (WRR_3), and TBR_4 (WRR_4). These new service classes are called Tag Bit Rate (TBR) classes. TVCs and ATM Forum VCs can only coexist on the same physical interface, but they SIN in the night mode and are unaware of each other.

TBR classes support only best-effort VCs (similar to the ATM Forum service category UBR); therefore, there is no bandwidth guarantee from the RS for TVCs. All of the TVCs fall into one of the four TBR classes, which each carry a different default relative weight. The default values of the relative weights for the four TBR classes are configurable so that you can change the priority of the default values.

Table 10-1 shows the TBR classes and ATM Forum class mappings into the service classes for physical ports.

TBR Class	Service Class	Relative Weight
TBR_1 (WRR_1)	1	1
TBR_2 (WRR_2)	6	2
TBR_3 (WRR_3)	7	3
TBR_4 (WRR_4)	8	4
ATM Forum Service Category	Service Class	Relative Weight
VBR-RT	2	15
VBR-NRT	3	2
ABR	4	2
UBR	5	2

Table 10-1: Service Class to Weight Mapping for Physical Ports

TBR Class Service Class Relative Weight

TBR_1 (WRR_1)

1

1

TBR_2 (WRR_2)

6

2

TBR_3 (WRR_3)

7

3

TBR_4 (WRR_4)

8

4

ATM Forum Service Category Service Class Relative Weight

VBR-RT

2

15

VBR-NRT

3

2

ABR

4

2

UBR

5

2

Note The CBR service category is mapped to service class 2, but all CBR VCs are rate scheduled only, and therefore are not WRR scheduled.

When tag switching is enabled on a hierarchical VP tunnel, the tunnel can only be used for tag switching. Because hierarchical VP tunnels support only four service classes, both TVCs and ATM Forum VCs map to the same service classes. Therefore, both ATM Forum VCs and TVCs cannot coexist in a hierarchical VP tunnel. The relative weights assigned to the service classes depend on which is active (either tag switching or ATM Forum). The class weights change whenever a hierarchical tunnel is toggled between ATM Forum and tag switching. By default, a hierarchical VP tunnel comes up as an ATM Forum port.

Table 10-2 lists the mapping of ATM Forum service categories and TBR classes for hierarchical VP tunnels.

TBR Class	Service Class	Relative Weight
TBR_1 (WRR_1)	1	1
TBR_2 (WRR_2)	2	2
TBR_3 (WRR_3)	3	3
TBR_4 (WRR_4)	4	4
ATM Forum Service Category
VBR-RT	1	15
VBR-NRT	2	2
ABR	3	2
UBR	4	2

Table 10-2: Service Class to Weight Mapping for Hierarchical VP Tunnels

TBR Class Service Class Relative Weight

TBR_1 (WRR_1)

1

1

TBR_2 (WRR_2)

2

2

TBR_3 (WRR_3)

3

3

TBR_4 (WRR_4)

4

4

ATM Forum Service Category

VBR-RT

1

15

VBR-NRT

2

2

ABR

3

2

UBR

4

2

Each service class is assigned a relative weight. These weights are configurable, and range from 1 to 15. Configuring the service class and relative weight requires the following steps:

Step 1 Select the interface to configure and enter interface configuration mode.

Step 2 Enter the service class and relative weight for the interface.

Threshold Group for TBR Classes

A threshold group utilizes the memory efficiently among VCs of a particular traffic type. Each threshold group is programmed with a dynamic memory allocation profile that maps into the needs of the connections of a particular service class.

The number of threshold groups available on the ATM switch router is platform dependent. For details, refer to the ATM Switch Router Software Configuration Guide.

Each threshold group has a set of eight regions, and each region has a set of thresholds. When these thresholds are exceeded, cells are dropped to maintain the integrity of the shared memory resource.

Each ATM Forum service category is mapped into a distinct threshold group. All the connections in a particular service category map into one threshold group. Similarly, all the TBR classes have best effort traffic; the service differentiation comes mainly by assigning different weights. Each of the TBR classes map into four different threshold groups whose parameters are the same as the UBR threshold group.

Table 10-3 shows the threshold group parameters mapped to the connections in all of the TBR classes for the ATM switch router.

Group	Maximum Cells	Maximum Queue Limit	Minimum Queue Limit	Mark Threshold	Discard Threshold	Use
7	131071	511	31	25%	87%	TBR_1 (WRR_1)
8	131071	511	31	25%	87%	TBR_2 (WRR_2)
9	131071	511	31	25%	87%	TBR_3 (WRR_3)
10	131071	511	31	25%	87%	TBR_3 (WRR_4)

Table 10-3: Threshold Group Parameters for TVCs

Group Maximum Cells Maximum Queue Limit Minimum Queue Limit Mark Threshold Discard Threshold Use

7

131071

511

31

25%

87%

TBR_1 (WRR_1)

8

131071

511

31

25%

87%

TBR_2 (WRR_2)

9

131071

511

31

25%

87%

TBR_3 (WRR_3)

10

131071

511

31

25%

87%

TBR_3 (WRR_4)

Each threshold group is divided into eight regions. Each region has a set of thresholds which are calculated from the corresponding threshold group parameters given in Table 10-3. The threshold group might be in any one of the regions depending on the fill level (cell occupancy) of that group. And that region is used to derive the set of thresholds which apply to all the connections in that group.

Table 10-4 gives the eight thresholds for threshold groups 6, 7, 8, and 9.

Table 10-4: Region Thresholds for Threshold Groups
Region Lower Limit Upper Limit Queue Limit Marking Threshold Discard Threshold

0

0

8191

511

127

447

1

8128

16383

255

63

223

2

16320

24575

127

31

111

3

24512

32767

63

15

63

4

32704

40959

31

15

31

5

40896

49151

31

15

31

6

49088

57343

31

15

31

7

57280

65535

31

15

31

**Table 10-4: Region Thresholds for Threshold Groups**
Region	Lower Limit	Upper Limit	Queue Limit	Marking Threshold	Discard Threshold
0	0	8191	511	127	447
1	8128	16383	255	63	223
2	16320	24575	127	31	111
3	24512	32767	63	15	63
4	32704	40959	31	15	31
5	40896	49151	31	15	31
6	49088	57343	31	15	31
7	57280	65535	31	15	31

For more information about threshold groups, see the "Threshold Groups" section in the chapter "Traffic and Resource Management."

CTT Rows

A row in the connection traffic table (CTT) is created for each unique combination of traffic parameters. When a TVC is set up in response to a request by tag switching, a CTT row is obtained from the resource manager by passing the traffic parameters which include the service category (TBR_x [WRR_x], where x is 1, 2, 3, or 4). If a match is found for the same set of traffic parameters, the row index is returned; otherwise, a new table is created and the row index of that CTT row is returned. Since all data TVCs use the same traffic parameters, the same CTT row can be used for all TVCs of a particular service category once it is created.

Note There are no user configurable parameters for the CTT with TVCs.

Resource Management CAC

Connection admission control (CAC) is not supported for TVCs. All TVCs are best effort connections; therefore, no bandwidth is guaranteed by the RS. Only the WRR scheduler is used. All of the traffic parameters (PCR, MCR, MBS, CDVT, and SCR) are unspecified. There is no best effort limit, as there is with ATM Forum UBR and ABR connections. CAC is bypassed for TVCs.

Table of Contents

Tag Switching