Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements
Webcast-Catalyst9k

ASK THE EXPERT- PPP & PPP MULTILINK

Welcome to the Cisco Networking Professionals Ask the Expert conversation. This is an opportunity to discuss PPP & PPP Multilink with Cisco expert Makarand Chitale. Makarand’s current responsibilities include Technical Assistance Center (TAC) Escalation in which he troubleshoots complex issues, provides training, and authors documentation. His areas of expertise are Configuration and Troubleshooting of Access Servers, PPP, VPDN etc. Feel free to post any questions relating to PPP & PPP Multilink. Remember to use the rating system to let Markarand know if you’ve received an adequate response.

Markarand might not be able to answer each question due to the volume expected during this event. Our moderators will post many of the unanswered questions in other discussion forums shortly after the event. This event lasts through August 15. Visit this forum often to view responses to your questions and the questions of other community members.

44 REPLIES
New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

We're about to try our first PPP Multilink, adding a second point-to-point T1 between a 3640 and 3660. Any tips on having minimal disruption to production traffic? We do need to install a CSU/DSU module in the 3640. IOS 12.1(3a)T1....unless we upgrade as part of this process.

Thank you for doing this forum. Good timing!

Paul Nicolette

NJ Dept of Health & Senior Services

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Just a few quick points on Multilink PPP, it's the load balancing techniques in that it sends packets across the individual links in a round robin fashion. However, Multilink PPP adds three significant capabilities.

1)Because multilane PPP works at the link layer, it makes an NxT1 bundle appear as one logical link to the upper layer protocols in the router. Thus, only one network address needs to be configured for the entire NxT1 bundle.

2)Multilink PPP keeps track of packet sequencing and buffers packets that arrive early. With this ability, Multilink PPP preserves packet order across the entire NxT1 bundle.

3)Packet fragmentation can be enabled to split large data packets into smaller packet fragments that are individually transmitted across the links. In many circumstances fragmentation can increase the efficiency of the NxT1 link.

Multilink PPP also offers significant link manageability. If an individual link fails, Multilink PPP can automatically detect the failure and remove that link from the bundle. As long as one link of the bundle is working the logical interface representing the bundle is up. By using keep alive signals, Multilink PPP can detect when a previously failed link comes back up then reinsert that link into the bundle. Additionally, when more bandwidth is needed, additional links can be added to the bundle by simply configuring them as a member of the bundle. No reconfiguration at the network layer, e.g., new addressing, is needed.

The tradeoff for the increased functionality is that Multilink PPP requires greater CPU processing than load balancing solutions. Packet reordering, fragment reassembly, and the Multilink PPP protocol itself all take CPU power.

Regarding service disruption when adding the second T1, it's important to know what the present configuration on the routers is, the ppp multilink is negotiated at the LCP stage so we might have to momentarily bring the first T1 link down by configuring "ppp multilink" command under the first T1 link. Also it's important to know whether we have authentication enabled or not, the bundling is done based on authenticated name or endpoint discriminator .

http://www.cisco.com/warp/public/131/mppp_bundle_name.html

http://www.cisco.com/en/US/tech/tk713/tk628/technologies_configuration_example09186a00800a5efa.shtml

http://www.cisco.com/en/US/tech/tk713/tk507/technologies_configuration_example09186a0080093d69.shtml

What card is being used for the addition T1 link?

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Mak,

Thank you for the helpful info. Card is a WIC 1DSU T1 (we're plugging in a 2FE 2W carrier). Authentication is disabled. Part of the config at the remote site appears below...let me know if you need more!

One other question - does Multichassis Multilink work across routers or just Access Servers?

- Paul

interface Serial0/0

description EXISTING T1 TO BE MULTILINKED

ip address 192.168.3.2 255.255.255.0

ip wccp 0 redirect out

no ip mroute-cache

ipx network xxxxxxxx

compress stac

service-module t1 remote-alarm-enable

!

interface FastEthernet0/1

description To web cache at 10.x.x.x

ip address 10.x.x.x 255.255.255.252

ip wccp redirect exclude in

duplex auto

speed auto

!

interface Serial0/1

description To another site - no multilink planned

ip address 192.168.10.1 255.255.255.252

no ip mroute-cache

compress stac

service-module t1 clock source internal

service-module t1 remote-alarm-enable

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Thanks for the inputs, the multilink can be configured using the below sample configs:

http://www.cisco.com/warp/public/131/7.html

Multichassis multilink can be configured on any routers, they need not be the Access servers.

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Thank you for the pointer to the helpful config.

One vendor SE I spoke with said that his testing with dual T1s found that load was balanced more evenly at Layer 3...w/EIGRP and CEF...than w/Multilink PPP, though packet sequence may not be preserved..

Do you agree? Would interleaving improve the link efficiency?

Also, is "compress stac" supported/recommended? We use it now. Thanks for helping give me a better understanding of the options and their implications.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Is it possible to use Multichassis Multilink PPP between two sites in a permanent fashion? I realize MMP is mostly for dial-in ISDN use. I have a site that has two T1s that connect into two different routers at my distro layer for redundancy. I'd like to be able to bundle those T1s together in a multilink to look like one 3 Mb/sec link. Is this possible? I'd also like to have a backup in case the head-end router that terminates the multilink bundle fails.

Thanks,

Chris McCoy

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Hi Chris,

Sure, it's possible to have Multichassis Multilink ppp between two sites on a permanent basis...In addition to ISDN , MMP is widely used for Async as well as for non-Dialer interfaces too.

http://www.cisco.com/warp/public/131/6.html#async

For your site that has 2 x T1s at one end on the same router connecting to 2 xT1s on different routers at the other end, will need to have these two routers in the same SGBP stack group & can keep the seed-bid to default or make one as the master always based on the platforms in use since in practice, an SGBP bid from a stack member is a function of locality, a user-configurable weighted metric, CPU type, number of MP bundles, and so on.....so basically we can choose which router (if different platforms) among the two is best suited.

http://www.cisco.com/warp/public/131/3.html

What kind of backup link do you have in mind? You say when the n\head end router fails...in that case we need another router to terminate the backup link.

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Mak,

Regarding the backup, I think I understand now that I have to use virtual-templates and not rotary groups. I tried using Dialer rotary-groups with SGBP a little while back. It seemed to work, but packets kept arriving out of order. Is a multilink bundle terminated with SGBP/L2TP have a different method for reordering packets compared to a multilink-group? Is it not as strict?

Also, regarding virtual templates, is there a way of using more than one virtual template configuration per destination, rather than one uniform virtual template for all MMP connections?

Thanks,

Chris M.

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Chris, I am not sure I get your question correctly when you say SGBP with backup, can you please explain a little more on what the intended setup will look like. As per my previous posting: What kind of backup link do you have in mind? You say when the head end router fails...in that case we need another router to terminate the backup link...can you please confirm the backup setup.

Although we can configure multiple virtual-template interfaces but can only have one multilink virtual-template x command globally, so basically with MMP we can only use one VTemp interface.

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

We are in the process of aggregating our wide area sites in to a central location. Our strategy has been to aggregrate T1's on to Cisco 7206's with channelized T3 cards using Telco Systems Edgelink 100 muxes. This gives us the ability to make changes quickly, multilink, and to add additonal circuits to sites for VOIP and video conferencing. Our test site has three T1's multilinked, but we are receiving a high number of CRC and input errors. While I understand there are lot's of variables that could cause the erros. Is there anything with the pa-mc-t3 cards, or in the setup of multilink that could cause the errors. I have individual T1's that are not multilinked that don't have errors. This particular test site has the 7200 at each end with T3 cards and the telco systems muxes.

Any input is appreciated.

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

I understand that you have a pa-mc-t3 connected to the mux with just a few T1 from the provider (not a complete T3), how are we configured the multilink here, using Virtual-template or using multilink interfaces? Also the crc errors are seen only after we configure the T1 for multilink...do we see them under " sh int mutilink x" or under "sh interface serial x"?

Is the T3 controller taking clock from line? Do we see any errors on the controller t3 & t1?

On a serial link, CRCs usually indicate noise, gain hits, or other transmission problems on the data link. Can you try changing the port on the mux.

The multilink config can be done as in the below URLs, this should not really induce crc errors on the interface... Also can you try disabling fragmentation on ppp multilink.

http://www.cisco.com/en/US/products/sw/iosswrel/ps1835/products_configuration_guide_chapter09186a00800ca724.html#1002432

http://www.cisco.com/en/US/tech/tk713/tk507/technologies_configuration_example09186a0080093d69.shtml

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Hello

Just a few general questions

What timeout values does multilink ppp use before it recognises a link in a ppp bundle has failed.

How many keepalives must it miss ?

What happens to the data passing down the failed link is it lost ?

How long does it take to add a failed link back into a ppp bundle.

Can these values be changed when considering a serial link i.e. T1/E1

Regards

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

We are experiencing an intermittent problem with Multilink over a PRI between two 3640s with 12.2(13)T5.

When the bundle is formed the minimum number of channels connect but the load is not distributed across the bundle with only 1 carrying traffic. This problem is only seen in one direction (outgoing from the end being instructed to call back the other direction is fine.

Direction placing return calls - not distributed

Direction receiving calls - OK

Example - chan 22 loaded and 2 not as seen from the receiving end

Dialer1 is up, line protocol is up (spoofing)

reliability 255/255, txload 66/255, rxload 26/255

Encapsulation PPP, LCP Open, multilink Open

Open: CDPCP, IPCP, loopback not set

Interface is bound to Se1/0:2

Interface is bound to Se1/0:4

Interface is bound to Se1/0:6

Interface is bound to Se1/0:8

Interface is bound to Se1/0:10

Interface is bound to Se1/0:19

Interface is bound to Se1/0:22

Interface is bound to Se1/0:25

Last input 00:00:04, output never, output hang never

Last clearing of "show interface" counters 3w0d

Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 63823

Queueing strategy: weighted fair

Output queue: 0/1000/64/63822/194 (size/max total/threshold/drops/interleaves)

Conversations 0/2/16 (active/max active/max total)

Reserved Conversations 1/1 (allocated/max allocated)

Available Bandwidth 0 kilobits/sec

30 second input rate 53000 bits/sec, 22 packets/sec

30 second output rate 134000 bits/sec, 70 packets/sec

11850771 packets input, 2447788052 bytes

13730966 packets output, 4139542507 bytes

Bound to:

Serial1/0:2 is up, line protocol is up

reliability 255/255, txload 71/255, rxload 1/255

Encapsulation PPP, LCP Open, multilink Open, loopback not set

Interface is bound to Di1 (Encapsulation PPP)

Last input 00:12:09, output 00:00:00, output hang never

5 minute input rate 0 bits/sec, 0 packets/sec

5 minute output rate 18000 bits/sec, 20 packets/sec

Bound to:

Serial1/0:22 is up, line protocol is up

Hardware is DSX1

MTU 1500 bytes, BW 64 Kbit, DLY 20000 usec,

reliability 255/255, txload 71/255, rxload 235/255

Encapsulation PPP, LCP Open, multilink Open, loopback not set

Interface is bound to Di1 (Encapsulation PPP)

Last input 00:00:00, output 00:00:00, output hang never

Last clearing of "show interface" counters never

Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 0

Queueing strategy: fifo

Output queue: 0/40 (size/max)

5 minute input rate 59000 bits/sec, 101 packets/sec

5 minute output rate 18000 bits/sec, 20 packets/sec

1015982 packets input, 72127231 bytes, 0 no buffer

Received 0 broadcasts, 0 runts, 0 giants, 0 throttles

0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored, 0 abort

1472528 packets output, 108619166 bytes, 0 underruns

0 output errors, 0 collisions, 169 interface resets

0 output buffer failures, 0 output buffers swapped out

168 carrier transitions

Timeslot(s) Used:23, Transmitter delay is 0 flags

interface Dialer1

ip unnumbered Loopback0

encapsulation ppp

ip route-cache flow

no ip mroute-cache

load-interval 30

dialer pool 1

dialer remote-name abababab

dialer idle-timeout 300 either

dialer fast-idle 5

dialer enable-timeout 30

dialer string 0740402600

dialer caller 740402600 callback

dialer hold-queue 20

dialer load-threshold 100 either

dialer max-call 16

dialer-group 2

service-policy output DEF

ppp authentication chap callin

ppp chap hostname cdcdcdcd

ppp multilink

ppp multilink fragment-delay 10

ppp multilink links maximum 16

ppp multilink links minimum 8

ppp multilink interleave

multilink load-threshold 100 either

thanks

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

When this problem happens can you please post the output of "sh ppp multilink" and the same outputs posted above, but from both ends..

Yes, we can see that at the receiving end the rxload is uneven but the txload is even...what about the other links in the bundle (s1/0:4, 6, 8 etc), do they too show uneven receive loads?

Serial1/0:2 is up, line protocol is up

reliability 255/255, txload 71/255, rxload 1/255

Serial1/0:22 is up, line protocol is up

reliability 255/255, txload 71/255, rxload 235/255

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Hi Mac,

I have configured cisco 805 router for small office to ISP.I created chat-script for this to work. But need some input from you:

Scenario: PC----(e0)Router(s0)---modem-----ISP

1)I can start successful dial-in session by revers telnet to s0( telnet 2001), it connects to ISP through modem. I can ping ISP's IP.

2)But cannot ping from my PC to ISP's IP, although I can ping to router's e0 from PC. IP route to Dialer interface( ip router 0.0.0.0 0.0.0.0 dialer 1).

I have worked with NAT before, in this very case, I did not use it because I put 2 valid IP addresses to routers e0 and PC's lan adapter.

e0-->202.168.254.241

PC-->202.168.254.243

But still, advise me, if I can use the following nat command to my router( from 805 doc):

"ip nat inside source list 1 interface dialer 0 overload"

3)This thing should be DDR, whenever my primary link should fail, any IP packet destained for internet should invoke the chat-script.How can I enable auto dial from my router if primary link should fail? Primary link resides in VPN box through Radio-Link to ISP.

Thanks in advance for your reply and help.

Richard

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

How does the S0 get an IP address, is the async interface (dialer 1) configured for "ip address nego"? OR is the dialer 1 unnumbered to e0? If using ip address negotiated what IP address range do we get from..."sh ip interface brief". Are these 202.168.254.xxx statical addresses assigned by the provider ? In that case the ISP needs to know where there addresses are located (looks like the ISP does not know where to send the packet when it's addressed for 202.168.254.xxx).

The way to test is by doing an extended ping to the ISP with the sourse ip address of e0 after the dialup connection is UP.

You said that you can dial using atdt (reverse telnet) & ping, it's going to be an exec session & you should not get any ip address...not sure how you can do this unless you get an router exec prompt & are on a router.

Please talk to the provider & ensure they have the correct routing statement on their Access server for the static addresses assigned to you.

Sure you can use NAT as well:

http://www.cisco.com/warp/public/471/ddreasyip.html

and should be dialer 1 overload in the statement below:

"ip nat inside source list 1 interface dialer 0 overload"

Regarding the backup, how is this setup? Is the Primary link & backup on the same router or different equipments?

For the c805 can you please send me the following:

sh run / sh ip int brief after the dialup connection is connected to the ISP.

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Good afternoon,

We have a couple of central 7206 routers with a dual T3 controllers (PA-MC-2T3+) running both channelized (T1s to remote sites) and unchannelized (45Mbits to our Data Center). I was wondering if you could tell me what ideally the settings should be for Framing, Line coding and Clocking. For example one of them has a framing of C-bit and another of M23. Others have Line Clocking or Internal Clocking. Not sure what is supposed to be at.

Thanks

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

These parameter configurations on the routers will mainly depend on what the mux or ADM (add drop mux for SONET) or switch is configured for. Basically we need to match them.

Regarding the clocking for WAN should be sourced from line.

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

This is true...... Its a general rule of thumb to match whatever the provider has set on the parameter build of the circuit. Framing and Line Coding SHOULD ALWAYS match what is set on the physical circuit....Clocking is usually set by Line for almost all providers unless otherwise specified during the time you initally order the circuit. A good way to tell is to review your error counts on "Show Interface", this is were you will normally be able to tell if any of these are incorrect or not. I hope this helps.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Dear Makarand,

We are investigating the possible development of a device that can provide multiple Managed VLAN ports at a remote site via an E1 link. Our device will be connected to either a Cisco or other Router through an E1 interface.

What we need to know is:

1. What functionality do we have to support on our device w.r.t. Ethernet switching/routing/bridging etc?

2. What protocols would we need to support on our device if PPP with HDLC like framing is used over the E1 link?

3. Would our product be similar to a Remote MAC Bridge or something completely different?

4. Are there any products available on the market that you know off that can provide the functionality we require?

5. Where can one get information on possible design architectures for this type of solution that we can use as a blue print for our design?

Can you help in providing answers to these questions.

Thanks in advance

Willie Engelbrecht

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Dear Makarand,

We have installations with 7500 routers terminating 64/128 kbps leased lines on PA-MC-8E1 with VoIP traffic. In such a situation everything (MLPPP, QoS, cRTP) must be done by the VIP cards in a distributed architecture.

As the number of lines increases the VIP cards become more and more instable. Is there any guideline about how many such lines can be served by a VIP? Also what parameters do you recommend to monitor?

Thanks for any suggestions,

Martin

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

When you say unstable do mean that the calls disconnect intermittently or you see issues with audio quality or more attempts required for call to connect etc etc.

After roughly how many calls do you see this issue appear? Will need to know the number of resources available on the 7500.

sh diag / sh ver will be useful with a brief description of what unstable means.

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Dear Makarand,

I mean unstable is the router's VIP itself. We're running 12.2(11)T8 because with later 12.2T releases we had much more problems. Even with this we have line flapping on MLPPP interfaces (the remote is a c805), and VIP crashes.

We have VIP2-40s.

Thanks,

Martin

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Hi Martin,

If we are seeing a VIP crash issue, I would recommend opening a TAC case with the following info attached for speedy resolution..

sh tech / sh context / the crash info file and syslog or router log messages.

For more info on troubleshooting crash issues:

http://www.cisco.com/en/US/products/sw/iosswrel/ps1835/products_tech_note09186a00800b4447.shtml

Thanks, Mak.

Anonymous
N/A

Re: ASK THE EXPERT- PPP & PPP MULTILINK

can you post more details about this issue

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Hello,

I have one question - it is a config issue. How to configure the delay before the isdn backup link will be dialed? In my scenario two sites are connected via serial and there is a ISDN backup link configured between them. The backup is configured as a dialer-map on physical interface on one site and as a dialer on the second. The backup is on floating statics. Both sites have dialer strings and have to be able to dial. I'd like to delay the dialing from one site after failure of the serial link. Is it possible?

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

In case of floating static routes, the static route (with the higher admin distance) is immediately inserted into the routing table as soon as the primary route (dynamically learn) is lost & then based on interesting traffic the call is initiated .....basically the option to delay the dialing process will require manipulation with the interesting traffic, for eg, say NTP will trigger the call. Mind you, there is a tradeoff in this since the idle timer will be controller by what traffic you configure as interesting.

Backup interface or Dialer watch can introduce delay, but floating is too simple for that level of customization.

http://www.cisco.com/warp/public/123/backup-main.html#floating_static_routes

http://www.cisco.com/pcgi-bin/Support/browse/psp_view.pl?p=Internetworking:DDR&viewall=true

Thanks, Mak.

New Member

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Hello Mak,

Thanks for the response. But what happens if we configure ppp multilink on both sides, both sites dial to each other almost simultaneusly after failure of the serial and establish two isdn calls on each router - one "in" and one "out". Can it be configured to negotiate the multilink on those both connections and drop one of them after some time? How do the routers agree which connection to drop (not to tear down both from both routers). And will the router learn to pass traffic to the B channel which was not establish by it and was left after disconnecting the other channel.

Thanks, Krzysztof

Silver

Re: ASK THE EXPERT- PPP & PPP MULTILINK

Hi Krzysztof,

In your scenario we are using floating static routes on either end router with both routers capable of dialing out. Since both these routers will detect the failed link (and insert the floating route) simultaneously & if both end routers have intersecting traffic at the same time they will to trigger the call to the other end coz they do not see any link UP to pass traffic....

NOTE, suppose side A initiated the call & got the first B channel up & after that side B needs to send traffic across then it will do so on the already existing link & will not bring up another link ..........until the load increases and based on the "dialer load-threshold" another link needs to be initiated.

Q) both sites dial to each other almost simultaneusly after failure of the serial and establish two isdn calls on each router - one "in" and one "out".

A) For a given router having one B channel as IN & the other as OUT does not really matter since they are both in the same bundle (virtual) & should be able to pass traffic just fine.

Q) Can it be configured to negotiate the multilink on those both connections and drop one of them after some time? How do the routers agree which connection to drop (not to tear down both from both routers).

A) Based on the load threshold configured & idle timer, the second B channel will drop (and maybe subsequently the first B channel)...after that drops & the load increases then depending on which side has the increased load of interesting traffic will initiate the link.

Based on the requirement we can change the "dialer load-threshold x IN / OUT / EITHER".....outbound is default, also the "ppp timeout muktilink add xx" can be used.

http://www.cisco.com/warp/public/131/mppp-ddr.html

Thanks, Mak.

651
Views
19
Helpful
44
Replies
CreatePlease to create content