|
|
This chapter describes how to configure Multichassis Multilink PPP. It includes the following main sections:
For a complete description of the MMP commands in this chapter, see the Cisco IOS Dial Services Command Reference publication. To locate documentation of other commands that appear in this chapter, use the command reference master index or search online.
Prior to Release 11.2, Cisco IOS supported Multilink PPP (MLP). Beginning with Release 11.2, Cisco IOS software also supports Multichassis Multilink PPP (MMP).
MMP, however, provides the additional capability for links to terminate at multiple routers with different remote addresses. MMP can also handle both analog and digital traffic.
This feature is intended for situations with large pools of dial-in users, where a single chassis cannot provide enough dial ports. This feature allows companies to provide a single dialup number to its users and to apply the same solution to analog and digital calls. This feature allows internet service providers (ISPs), for example, to allocate a single ISDN rotary number to several ISDN PRIs across several routers. This capability allows for easy expansion and scalability, and for assured fault tolerance and redundancy.
MMP allows network access servers to be stacked together and appear as a single network access server chassis so that if one network access server fails, another network access server in the stack can accept calls.
With large-scale dial-out, these features are available for both outgoing and incoming calls.
Routers or access servers are configured to belong to groups of peers, called stack groups. All members of the stack group are peers; stack groups do not need a permanent lead router. Any stack group member can answer calls coming from a single access number, which is usually an ISDN PRI hunt group. Calls can come in from remote user devices, such as routers, modems, ISDN terminal adapters, or PC cards.
If a more powerful router is available, it can be configured as a member of the stack group and the other stack group members can establish tunnels and forward all calls to it. In such a case, the other stack group members are just answering calls and forwarding traffic to the more powerful offload router.
 |
Note High-latency WAN lines between stack group members can make stack group operation inefficient. |
MMP call handling, bidding, and Layer 2 forwarding operations in the stack group proceed as follows:
1. When the first call comes in to the stack group, router A answers.
2. In the bidding, router A wins because it already has the call. Router A becomes the call-master for that session with the remote device. (Router A might also be called the host to the master bundle interface.)
3. When the remote device that initiated the call needs more bandwidth, it makes a second MLP call to the group.
4. When the second call comes in, router D answers it and informs the stack group. Router A wins the bidding because it already is handling the session with that remote device.
5. Router D establishes a tunnel to router A, and forwards the raw PPP data to router A.
6. Router A reassembles and resequences the packets.
7. If more calls come in to router D and they too belong to router A, the tunnel between A and D enlarges to handle the added traffic. Router D does not establish an additional tunnel to A.
8. If more calls come in and are answered by any other router, that router also establishes a tunnel to A and forwards the raw PPP data.
9. The reassembled data is passed on the corporate network as if it had all come through one physical link.
In contrast to the previous figure, Figure 5 features an offload router. Access servers that belong to a stack group answer calls, establish tunnels, and forward calls to a Cisco 4700 router that wins the bidding and is the call master for all the calls. The Cisco 4700 reassembles and resequences all the packets coming in through the stack group.
|
Note You can build stack groups using different access server, switching, and router platforms. However, universal access servers such as the Cisco AS5200 should not be combined with ISDN-only access servers such as the Cisco 4000-series platform. Because calls from the central office are allocated in an arbitrary way, this combination could result in an analog call being delivered to a digital-only access server, which would not be able to handle the call. |
MMP support on a group of routers requires that each router be configured to support the following:
MMP does not require reconfiguration of telephone company switches.
To configure MMP, perform the tasks in the following sections, in the order listed:
See the section "Monitoring and Maintaining MMP Virtual Interfaces" later in this chapter for tips on maintaining MMP. See the examples in the section "MMP Configuration Examples" later in this chapter for ideas on how to configure MMP in your network.
To configure the stack group on the router, use the following commands beginning in global configuration mode:
| Command | Purpose | |
|---|---|---|
Step 1 | sgbp group name | Creates the stack group and assign this router to it. |
Step2 | sgbp member peer-name [peer-ip-address] | Specifies a peer member of the stack group. Repeat this step for each additional stack group peer. |
|
NoteOnly one stack group can be configured per access server or router. |
You need to configure a virtual template for MMP when asynchronous or synchronous serial interfaces are used, but dialers are not defined. When dialers are configured on the physical interfaces, do not specify a virtual template interface.
To configure a virtual template for any nondialer interfaces, use the following commands beginning in global configuration mode:
| Command | Purpose | |
|---|---|---|
Step1 | multilink virtual-template number | Defines a virtual template for the stack group.
|
Step2 | ip local pool default ip-address | Specifies an IP address pool by using any pooling mechanism---for example, IP local pooling or Dynamic Host Configuration Protocol (DHCP) pooling. |
Step3 | interface virtual-template number | Creates a virtual template interface, and enter interface configuration mode. |
Step4 | ip unnumbered ethernet 0 | Specifies unnumbered IP. |
Step5 | no ip route-cache | Disables fast switching, which enables per-packet load sharing and enhances performance on slower serial links. |
Step6 | encapsulation ppp | Enables PPP encapsulation on the virtual template interface. |
Step7 | ppp multilink | Enables Multilink PPP on the virtual template interface. |
Step8 | ppp authentication chap | Enables PPP authentication on the virtual template interface. |
If dialers are or will be configured on the physical interfaces, the ip unnumbered command, mentioned in Step 4, will be used in configuring the dialer interface. For examples that show MMP configured with and without dialers, see the "MMP Configuration Examples" at the end of this chapter.
To monitor and maintain virtual interfaces, use any of the following commands in EXEC mode:
| Command | Purpose |
|---|---|
show ppp multilink | Displays MLP and MMP bundle information. |
show sgbp | Displays the status of the stack group members. |
show sgbp queries | Displays the current seed bid value. |
The following examples show MMP configuration without and with dialers:
The following example shows the configuration of MMP when no dialers are involved. Comments in the configuration discuss the commands. Variations are shown for a Cisco AS5200 access server or Cisco4000 series router, and for an E1 or T1 controller.
sgbp group stackq sgbp member systemb 1.1.1.2 sgbp member systemc 1.1.1.3 username stackq password therock ! First make sure the multilink virtual template number is defined globally on ¡ each router that is a member of the stack group. multilink virtual-template 1 ! If you have not configured any dialer interfaces for the physical interfaces in ! question (PRI, BRI, async, sync serial), you can define a virtual template. interface virtual-template 1 ip unnumbered e0 no ip route-cache ppp authentication chap ppp multilink ! Never define a specific IP address on the virtual template because projected
! virtual access interfaces are always cloned from the virtual template interface.
! If a subsequent PPP link also gets projected to a stack member with a virtual
! access interface already cloned and active, we will have identical IP addresses
! on the two virtual interfaces. IP will erroneously route between them. ! On an AS5200 or 4XXX platform: ! On a TI controller ! controller T1 0 framing esf linecode b8zs pri-group timeslots 1-24 ! interface Serial 0:23 no ip address encapsulation ppp no ip route-cache ppp authentication chap ppp multilink ! ! On an E1 Controller ! controller E1 0 framing crc4 linecode hdb3 pri-group timeslots 1-31 interface Serial 0:15 no ip address encapsulation ppp no ip route-cache ppp authentication chap ppp multilink
When dialers are configured on the physical interfaces and when the interface itself is a dialer, do not specify a virtual template interface. For dialers, you only need to define the stack group name, common password, and its members across all the stack members. No virtual template interface is defined at all.
Only the PPP commands in dialer interface configuration are applied to the bundle interface. Subsequent projected PPP links are also cloned with the PPP commands from the dialer interface.
This section includes the following examples:
The following example includes a dialer that is explicitly specified by the interface dialer command and configured by the commands that immediately follow:
sgbp group stackq sgbp member systemb 1.1.1.2 sgbp member systemc 1.1.1.3 username stackq password therock interface dialer 1 ip unnumbered e0 dialer map ..... encapsulation ppp ppp authentication chap dialer-group 1 ppp multilink ! ! on a T1 controller ! controller T1 0 framing esf linecode b8zs pri-group timeslots 1-24 interface Serial0:23 no ip address encapsulation ppp dialer in-band dialer rotary 1 dialer-group 1 ! ! or on an E1 Controller ! controller E1 0 framing crc4 linecode hdb3 pri-group timeslots 1-31 interface Serial0:15 no ip address encapsulation ppp no ip route-cache ppp authentication chap ppp multilink
ISDN PRIs and BRIs by default are dialer interfaces. That is, a PRI configured without an explicit interface dialer command is still a dialer interface. The following example configures ISDN PRI. The D-channel configuration on serial interface 0:23 is applied to all the B channels. MMP is enabled, but no virtual interface template needs to be defined.
sgbp group stackq sgbp member systemb 1.1.1.2 sgbp member systemc 1.1.1.3 username stackq password therock isdn switch-type primary-4ess controller t1 0 framing esf linecode b8zs pri-group timeslots 1-23 isdn switch-type basic-net3 interface Serial0:23 ip unnumbered e0 dialer map ..... encap ppp ppp authentication chap dialer-group 1 dialer rot 1 ! ppp multilink
The following example shows a virtual template interface for a system being configured as an offload server (via the sgbp seed-bid offload command). All other stack group members must be defined with the sgbp seed-bid default command (or if you do not enter any sgbp seed-bid command, it defaults to this command).
multilink virtual-template 1 sgbp group stackq sgbp member systemb 1.1.1.2 sgbp member systemc 1.1.1.3 sgbp seed-bid offload username stackq password therock interface virtual-template 1 ip unnumbered e0 no ip route-cache ppp authentication chap ppp multilink
Posted: Tue Jul 18 15:05:26 PDT 2000
Copyright 1989-2000©Cisco Systems Inc.