getXML('MPLS2959 devang_etcomSep 17, 2009, 2:18pm PSTHi,<br /><br />is multicast is supported on Multi-vrf configured router? As I have two document where they are talking about the Multicast can be run on multi-vrf interface not running MPLS. Other doc is saying you can not run mulicast on interface running MPLS with multi-vrf cofniguration...<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/docs/ios/12_2sb/12_2sba/feature/guide/vrflite.html')">http://www.cisco.com/en/US/docs/ios/12_2sb/12_2sba/feature/guide/vrflite.html</A> <--- look at the restriction.<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/technologies/tk648/tk828/tk363/technologies_white_paper0900aecd8012033f.html')">http://www.cisco.com/en/US/technologies/tk648/tk828/tk363/technologies_white_paper0900aecd8012033f.html</A><br /><br />Why is it not possible to run the MULTICAST on VRF LITE interface running with MPLS on it?<br /><br />thanks,<br />Devang 30 hritterSep 17, 2009, 5:07pm PSTDevang,<br /><br />I am not too sure what that second is referring to. Multicast is certainly supported in a multi-vrf (also know as vrf-lite) context as multicast is VRF aware.<br /><br />Regards devang_etcomSep 17, 2009, 5:16pm PSTHritter,<br /><br />Thanks for your reply. The issue is will multicast work if we have MPLS between PE-CE link and CE is multi-VRF CE? According to me it will not... <br /><br />AS well as I am trying to find out an application where we have MPLS running on PE-CE link and CE is multi-VRF interface (Interface facing to PE on CE router is also part of VRF)... any idea?<br /><br />thanks,<br />Devang hritterSep 17, 2009, 6:08pm PSTDevang,<br /><br /><br />Why would you have MPLS on the PE-CE link? This is only required in certain corner cases (Carrier supporting Carrier for instance). <br /><br />Multicast will work on the CE even if this CE is configured for multi-VRF.<br /><br />Regards<br /> devang_etcomSep 17, 2009, 6:51pm PSTHritter,<br /><br />I trying to understand the same thing from the first link i posted in my first post. In that Cisco doc they are talking about MPLS between PE-CE link and also saying that IP Multicast will not be supported in restriction portion of the document. So there I have big confusion. I know Multicast works in PE-CE with multi-VRF CE but there it is saying IP Multicasting will not be supported when there is MPLS running on link between PE-CE link and CE is multi-vrf interface running VRF on interface facing to PE.<br /><br />Thanks,<br />Devang amit.bhagatSep 17, 2009, 5:32pm PSTHi All,<br /><br />This Cisco document (the first one) suggests that in VRF-Lite, MPLS labels are exchanged between PE and CE. However, most books suggest no labels are exchanged between PE and CE, and no MBGP is required.<br /><br />Here's an sample of the one I tested recently-<br /><A HREF="javascript:newWin('http://sites.google.com/site/amitsciscozone/home/mpls/multi-vrf-or-vrf')">http://sites.google.com/site/amitsciscozone/home/mpls/multi-vrf-or-vrf</A><br /><br />Is it important to run LDP/MBGP between PE and CE in VRF-Lite? What advantages you get if you run LDP/MBGP instead of not running one?<br /><br />Please share your thoughts.<br /><br />Amit. devang_etcomSep 17, 2009, 5:37pm PSTAmit,<br /><br />According to me adding MPLS between PE-CE link with CE acting as Multi-VRF CE (PE facing interface is also in VRF) will not support the Multicast... then you will have to add full mesh of GRE between required sites to support the multicast...<br /><br />even i am also looking for what are the advantages of running LDP/BGP on PE-CE link... if there is a case for CSC then its fine to have MPLS between PE-CE link but again MultiVRF CE with PE facing VRF interface will have issue with Multicast scalability...<br /><br />waiting for an replies from different engineers to get clear view of it...<br /><br />thanks,<br />Devang ulatifNov 17, 2009, 1:55am PSTMulticast should work in both of the following scenarios:<br /><br />1. CE doing VRF-Lite with no MPLS between PE-CE (assuming PE is doing MPLS towards core).<br /><br />2. PE-CE doing MPLS i.e. CsC<br /><br />Have you lab tested and found that this doesnt work? <br />IMO both of above should work. amit.bhagatSep 12, 2009, 4:13pm PSTHi All,<br /><br />I tried the following simple topology for mVPN-<br /><br />CE1-PE1-PE3-CE2<br /><br />CE1 is connected to the source and CE2 has the receiver.<br /><br />PE1 & PE3 VRF configuration-<br />ip vrf CUST1<br /> rd 1:1<br /> route-target both 1:1<br /> bgp next-hop Loopback 0<br /> mdt default 239.192.10.1<br /> mdt data 239.192.20.0 0.0.0.15 threshold 1 list 100<br />!<br />access-list 100 permit ip any any<br />!<br /><br />I could send the multicast traffic from one end to another using the Default-MDT. PIM adjacency is formed over the MTI Tunnel 0.<br /><br />The issue is- no matter how much multicast traffic is sent across (multicast traffic here is large ping packets), PE3 does not switchover to Data-MDT.<br /><br />Please share your thoughts.<br /><br />Amit. 30 laaubertSep 13, 2009, 2:53am PSTHi Amit,<br /><br />You don't need the ACL in your case and mdt-data cmd is only required on the PE connected to the source but it can't explain why it's not working.<br /><br />This is the PIM debug you should have on PE connected to the source:<br /><br />*Sep 13 10:36:01.683: PIM(1): MDT threshold exceeded for (40.1.1.1,224.1.1.1)<br />*Sep 13 10:36:01.683: PIM(1): MDT join sent for (40.1.1.1,224.1.1.1) MDT:239.0.0.128 Tunnel0<br /><br />you could try with a threshold of 5 and see if it change something.<br /><br />If your source is a router sending ping, please set the timeout to 0<br /><br />HTH<br /><br />Laurent. amit.bhagatSep 13, 2009, 4:09am PSTHi Laurent,<br /><br />The "debug ip pim" command should show Data-join TLVs from the PE router when the traffic for (S,G) entry exceeds the set threshold. But I didn't see any such messages.<br /><br />Just wondering, did you try this in ur lab? Which routers and IOS did you use? I used 2691 with 12.4(25a) IOS. According to Cisco feature navigator, it supports mVPN.<br /><br />Regards,<br />Amit. laaubertSep 13, 2009, 4:19am PSTYes, I tested it with 12.2(33)SRC for 7600 and 12.2(33)SXH for 6500.<br /><br />Laurent. ulatifSep 17, 2009, 4:29pm PSTCould you paste the output of following from your configuration on PE1 and PE3:<br /><br />"show run | i pim"<br />"show ip pim mdt bgp"<br /><br /><br />thanks<br />Usman amit.bhagatSep 17, 2009, 4:42pm PSTHi Usman,<br /><br />show ip pim mdt bgp output is as follows-<br /><br />PE1# show ip pim mdt bgp<br />Peer (Route Distinguisher + IPv4) Next Hop<br /> MDT group 239.192.10.1<br /> 2:1:1:3.3.3.3 3.3.3.3<br /><br /><br />Here's another output- It clearly states Default-MDT group is created.<br /><br />PE1#show ip pim mdt<br /> * implies group is the MDT default group<br /> MDT Group Interface Source VRF<br />* 239.192.10.1 Tunnel0 Loopback0 CUST1<br /><br />I dont have the configs at this stage,only had this output handy.<br /><br />I still need to try Laurent's tip of setting the timeout to 0 since I am send ping packets from a router.<br /><br />Will update this post once I implement that. shivluSep 17, 2009, 10:18pm PSTHi Amit<br /><br />The IOS you are using is SRC series and in the outputs you are getting extended community of 2:X:X. <br />Please let us know have you enabled the ipv4 mdt under bgp or not. <br /><br />address-family ipv4 mdt<br /> neighbor <ip address> send-community both<br /> neighbor <ip address> activate<br />exit-address-family<br /><br />regards<br />shivlu jain<br /><A HREF="javascript:newWin('http://shivlu.blogspot.com')">http://shivlu.blogspot.com</A> amit.bhagatSep 18, 2009, 6:08am PSTHi Shivlu,<br /><br />I have used this IOS on 2691 router-<br />c2691-spservicesk0-mz.124-25a.bin<br /><br />Also, I did not enable ipv4 mdt. Most books do not have it enabled, however they could be using older IOS then.<br /><br />I might have to try enabling ipv4 mdt.<br /><br />Regards,<br />Amit. amit.bhagatNov 16, 2009, 3:20pm PSTHi All,<br /><br />Just like to provide an update that this issue was resolved when I switched over to 7200 series routers. I was able to switchover to the Data-MDT as soon as the threshold was reached.<br /><br />Thank you all.<br /><br />Regards,<br />Amit. ulatifNov 16, 2009, 9:28pm PSTI once had a similar issue in my network where a CPP (control plane policing) policy was dropping the data-MDT joins.<br />Not saying this is the reason in your case but you might wanna check on these lines:<br /><br />i.e. whether or not the data MDT join is being generated and makes it through the network while using the 7600/6500asaykao73Nov 16, 2009, 5:31pm PSTHI All,<br /><br />I'm a bit lost as to why our mpls network is broken between our POPS. I don' know why I can't ping from PE2(lo99) to PE1(lo99). Because of this I can't establish a MP-BGP session between PE1 and PE2. <br /><br />POP1[PE1 (lo99:172.16.99.13) -> P1] -> POP2[P2 -> PE2 (lo99: 172.16.99.4)]<br /><br />PE2 can ping P1.<br />P2 can ping P1.<br />P2 and PE2 CAN NOT ping PE1.<br />VPN traffic seems to not get to P1 and go further.<br /><br />PE2#ping 172.16.99.13<br />Type escape sequence to abort.<br />Sending 5, 100-byte ICMP Echos to 172.16.99.13, timeout is 2 seconds:<br />.....<br />Success rate is 0 percent (0/5)<br /><br />PE2#traceroute mpls ipv4 172.16.99.13 255.255.255.255<br />Tracing MPLS Label Switched Path to 172.16.99.13/32, timeout is 2 seconds<br /> <br />Type escape sequence to abort.<br /> 0 203.10.110.207 MRU 1500 [Labels: 3034 Exp: 0] <-- PE2<br />R 1 203.10.110.211 MRU 9000 [Labels: 8932 Exp: 0] 184 ms <-- packet reaches P2<br />. 2 * <-- this next hop should be P1 but packet not getting there<br />. 3 *<br />. 4 *<br /><br />* It was working fine til we switched our P1>P2 link to switched ethernet. Our upstream<br />provider says everything is ok on their end and that jumbo frames are enabled within their<br />switched ethernet network.<br /><br />* I've traced the labels from PE2 back to P2 > P1 > PE1 and they look ok.<br /><br />PE2#sh mpls forwarding-table 172.16.99.13 32<br />Local Outgoing Prefix Bytes Label Outgoing Next Hop<br />Label Label or VC or Tunnel Id Switched interface<br />617 3034 172.16.99.13/32 0 Gi0/0.11 203.10.110.211<br /><br />P2#sh mpls forwarding-table 172.16.99.13 32<br />Local Outgoing Prefix Bytes Label Outgoing Next Hop<br />Label Label or VC or Tunnel Id Switched interface<br />3034 8668 172.16.99.13/32 1582342 Gi4/0/1 203.17.96.97<br /><br />P1#sh mpls forwarding-table 172.16.99.13 32<br />Local Outgoing Prefix Bytes Label Outgoing Next Hop<br />Label Label or VC or Tunnel Id Switched interface<br />8668 Pop Label 172.16.99.13/32 158253163 Gi0/0.152 203.17.102.113<br /><br />* Odd thing is that I can ping from PE1 > PE2 and PE1 > P2. <br />Just can't ping it the other way P2 > PE1 and PE2 > PE1.<br /><br />PE1#ping 172.16.99.4<br />Type escape sequence to abort.<br />Sending 5, 100-byte ICMP Echos to 172.16.99.4, timeout is 2 seconds:<br />!!!!!<br />Success rate is 100 percent (5/5), round-trip min/avg/max = 8/10/12 ms<br /><br /><br />* I did a 'debug mpls packet' on P1 and can see labels being swapped ok (I think).<br /><br />Nov 16 22:43:24.546 AEDT: MPLS turbo: Gi0/1: rx: Len 78 Stack {8668 6 255} - ipv4 data<br />Nov 16 22:43:25.606 AEDT: MPLS turbo: Gi0/0.152: rx: Len 86 Stack {7087 6 255} - ipv4 data<br />Nov 16 22:43:25.606 AEDT: MPLS turbo: Gi0/2: tx: Len 82 Stack {6039 6 254} - ipv4 data<br />Nov 16 22:43:26.878 AEDT: MPLS turbo: Gi0/0.152: rx: Len 86 Stack {7203 6 255} - ipv4 data<br />Nov 16 22:43:26.878 AEDT: MPLS turbo: Gi0/2: tx: Len 82 Stack {6628 6 254} - ipv4 data<br />Nov 16 22:43:27.358 AEDT: MPLS turbo: Gi0/0.152: rx: Len 86 Stack {7647 6 255} - ipv4 data<br />Nov 16 22:43:27.362 AEDT: MPLS turbo: Gi0/2: tx: Len 82 Stack {728 6 254} - ipv4 data<br /><br /><br />* Any ideas as to why all of a sudden I can not longer ping from P2 > PE1 and PE2 > PE1? It looks like a tranmission issue somewhere between P1 and P2 because I can't pass labels beyond P1 as seen from the traceroute mpls output above. <br /><br />Please help...<br /><br />Thanks.<br /><br />Andy<br /> 30sharifimrNov 16, 2009, 7:31pm PSTHi Andy,<br /><br />0 203.10.110.207 MRU 1500 [Labels: 3034 Exp: 0] <-- PE2 <br />R 1 203.10.110.211 MRU 9000 [Labels: 8932 Exp: 0] 184 ms<br /><br />Do you have the same MTU configured on all your devices?<br /><br />Reza<br /><br />asaykao73Nov 16, 2009, 7:36pm PSTThe main link from one POP to another is configured with MTU 9000 (see below). And I don't usually touch the MTU between the P and PE routers which would default to 1500. <br /><br />It's always worked this way until we moved over to this new Switch Ethernet circuit for our backhaul from one POP to another. All of a sudden PE devices at POP2 can not ping PE devices at POP1 - BUT strange enough PE devices at POP1 can ping PE devices at POP2.<br /><br />--------------------------<br />P1 Config<br />--------------------------<br />interface GigabitEthernet0/2<br /> description Connection to P2<br /> mtu 9000<br /> bandwidth 150000<br /> ip address 203.17.96.x 255.255.255.252<br /> load-interval 30<br /> media-type gbic<br /> speed auto<br /> duplex auto<br /> negotiation auto<br /> mpls ip<br /><br />--------------------------<br />P2 Config<br />--------------------------<br />interface GigabitEthernet4/0/1<br /> description Connection to P1<br /> mtu 9000<br /> bandwidth 150000<br /> ip address 203.17.96.x 255.255.255.252<br /> load-interval 30<br /> negotiation auto<br /> mpls ip<br /><br />asaykao73Nov 16, 2009, 9:19pm PSTThis has been resolved.<br /><br />Turns out it was something within our Provider's network which does the backhaul for us that had some mac-access group configured on their switch which was blocking the PE's loopback from communicating with each other. maanyagoelOct 15, 2009, 10:10am PSTHello Everybody,<br /><br />I have a question regarding Auto-route announce feature. <br /><br />If Auto-route feature is enabled to advertise the subnets in a Tunnel use it in SPF calcualtion, then will I be able to use PIM protocol on the global network( on which Auto-route announce is enabled? <br /><br /><br /> 30 hritterOct 16, 2009, 11:42am PSTMaanya,<br /><br />This is correct but you will need the following command under the ospf or isis process to allow the reverse path forwarding to succeed when the route back to the source points at the tunnel interface:<br /><br />mpls traffic-eng multicast-intact <br /><br />Please refer to the following url for more information about this command:<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/docs/ios/12_0s/feature/guide/mplstemi.html')">http://www.cisco.com/en/US/docs/ios/12_0s/feature/guide/mplstemi.html</A><br /><br />Regards ulatifNov 16, 2009, 2:55pm PSTHarold,<br /><br />If the tunnel is only one-way ? do you still need this command ? :)<br /><br />e.g. lets say we have:<br /><br />Src----PE1----P1----PE2---Receiver<br /><br />If there is only one TE tunnel configured from PE1 -> PE2 but on the way back from PE2 to PE1 there is no tunnel?<br />I would think that you would not need this command in this scenario ?<br /><br />So multicast is not LS like unicast is and therefore if the route back to source points to tunnel interface, the command is needed to make RPF successful. syjeonSep 7, 2009, 3:42pm PST!<br />interface Tunnel1211<br /> ip unnumbered Loopback0<br /> load-interval 30<br /> tunnel destination 1.1.1.1<br /> tunnel mode mpls traffic-eng<br /> tunnel mpls traffic-eng autoroute announce<br /> tunnel mpls traffic-eng priority 1 1<br /> tunnel mpls traffic-eng bandwidth 1000<br /> tunnel mpls traffic-eng path-option 1 explicit name hj<br /> tunnel mpls traffic-eng path-option 2 dynamic<br /> tunnel mpls traffic-eng fast-reroute<br />!<br /><br />mpls traffic-eng reoptimize timers frequency 5<br />mpls traffic-eng reoptimize timers delay installation 5<br />mpls traffic-eng reoptimize timers delay cleanup 5<br />mpls traffic-eng reoptimize events link-up<br /><br /><br />Our mpls configuration is like above log.<br /><br />We configured the path option1, path option2 in the tunnel interface.<br />when the explicit path failed, the path option2 comes up.<br />Our plan is that when the path option 1 is failed, and the core running the FRR<br />We do not want to switch over th path option2 dynamically<br />How can I do that? Any thought?<br /><br />and next...<br />there is mpls traffic-eng reoptimize in our configuration. If I remove that, Can I make it like above?<br /> 30 jcozzupoliSep 7, 2009, 4:46pm PSTI think the best way is to have 2 TE GRE Tunnels configured. Both with explicit paths with one the direct path and the other one having the backup path. See below:<br /><br />interface Tunnel100<br /> description DIRECT LSP<br /> ip unnumbered Loopback0<br /> tunnel destination 1.1.1.1<br /> tunnel mode mpls traffic-eng<br /> tunnel mpls traffic-eng autoroute announce<br /> tunnel mpls traffic-eng path-option 1 explicit name TE_PRIMARY<br /> tunnel mpls traffic-eng priority 1 1<br /> tunnel mpls traffic-eng bandwidth 1000<br /> tunnel mpls traffic-eng fast-reroute<br /><br />interface Tunnel200<br /> description BACKUP LSP<br /> ip unnumbered Loopback0<br /> tunnel destination 1.1.1.1<br /> tunnel mode mpls traffic-eng<br /> tunnel mpls traffic-eng autoroute announce<br /> tunnel mpls traffic-eng path-option 1 explicit name TE_SECONDARY<br /> tunnel mpls traffic-eng priority 1 1<br /> tunnel mpls traffic-eng bandwidth 1000<br /><br />interface X<br /> mpls traffic-eng backup-path Tunnel200<br /><br />-----<br /><br />Also, depending on your IGP (IS-IS is faster for FRR convergence) you could tweak the spf timers and if you are running BFD then tie RSVP to BFD for even better convergence (but this depends on hardware platforms used). syjeonSep 7, 2009, 5:22pm PSTThank a lot..<br /><br />I understood your comment.<br />but I have a question within your config.<br />the interface X...<br />When We are config explcit path, we point hop-by-hop. <br />I think that your interface x is Physical interface. Right?<br /><br />if so. if the interface x didn't goes to down, the backup-path tunnel 200 didn't activate..Right? the reason is the physical link still up.<br /><br />if I can configure to make backup tunnel in the the interface tunnel 100, It's make sense. but real didn't do that.<br /><br /> jcozzupoliSep 7, 2009, 5:42pm PSTYeah interface X is the physical interface where your LSP begins. Remember, in the routing table you'll see the tunnels and not the physical interface. I have implemented this for a customer and it works pretty really well (sub 2ms failover).<br /><br />Another note, if you remove "autoroute announce" from the backup tunnel it will work only as a backup, if you have "autoroute announce" there it will load-share between the 2 tunnels.<br /><br />I recommend doing this in a lab environment so you can break the direct link and see the routing change. An MPLS traceroute or a normal IP traceroute will show you which LSP its taking (ie, which Tunnel is in use).<br /><br />HTH.<br /><br />- Joe. syjeonSep 7, 2009, 5:56pm PSTThans a lot! Be always happy :)<br /><br />I would like to know when the below,<br /><br />The Tunnel 100 is configured with Explicit Path. and It's working well in normal . but Tunnel 100 is down due to its explicit path failed. but Tunnel 200 didn't working due to tunnel 100 indirect link failed.<br /><br />AS I know. the physcal link still comes up. and the indirect explicit failed . so the traffic can't use the tunnel 200<br /><br />just thing I would like to know that If I want to activate the tunnel 200, the physical link must be down ? Right? jcozzupoliSep 7, 2009, 6:06pm PSTWell yes and no. Because like I said in my previous post, you can load-share across the 2 tunnels so this way you don't have any wasted bandwidth. This is done with my configuration having "autoroute announce" in both tunnels. jcozzupoliSep 9, 2009, 1:05am PSTHas your question been answer sufficiently? If so, can you please vote and advise Author's question has been answered?<br /><br />Thanks.<br />Joe. ulatifNov 16, 2009, 5:19am PSTOkay so you asked two questions:<br /><br />#1. If you have 2nd path option configured and if there is a failure on 1st path option, you would like the 1st path option to remain in place (using FRR) and 2nd path option should not kick in? is that a correct understanding of your question?<br /><br />If yes, then I'd have to ask you a question in return - why then you have a second path option configured if you want the TE tunnel to continue using FRR path option ?<br />What normally happens is that when the PLR detects the failure, it also sends an RSVP message to the head-end of the tunnel informing it about the failure in the path. This is why you normally would see that when FRR is in place for the 1st path option, you see tunnel stating "Change in resources detected: reroute pending" if you do a "show mpls traffic-eng tunnel" on the head-end.<br />Now if the head-end has a 2nd path-option configured, it would switch to the 2nd path-option (as it is supposed to do) in a make-before-break fashion. If you dont want it, then remove the 2nd path-option :)<br /><br />#2: Reoptimization timers - i dont think reoptimization timers will help you in achieving what i mentioned in part #1 because reoptimization is a scheduled event that has little to do with FRR and subsequent second path option kicking in.<br />It would help you more when your failed 1st path recovers and traffic needs to be switched back to 1st path option.<br /><br />regards<br />Usman jcozzupoliNov 16, 2009, 5:52am PSTIndeed Usman! :) hunter.lai@netstarnetworks.comOct 27, 2009, 9:48pm PSTany idea about 7600 MPLS interoperability with ALU 7750/7450? 30 ulatifNov 16, 2009, 5:28am PSTI have done some testing on the MPLS interoperatibility between ALC-7750 and Cisco 7600 and they seem to work fine.<br />Are you after a specific implementation?<br />I have tested the below:<br /><br />1. EoMPLS<br /><br />For this, the Alcatel implements it as an EPIPE service using an SDP which in Cisco is simple a targetted LDP session to the ALC. Rest is simple<br /><br /><br />MPLS L3VPNs<br />For this, the Alcatel uses what is called VPRNs using same BGP VPNv4 exchange and RD/RT concepts as used by Cisco.<br /><br />regards<br />UsmanOmahaGTP1Nov 15, 2009, 9:48am PSTJust a quick question.<br /><br />Currently I know of a command on an AdTran that would allow you to create a QoS policy and apply a 'traffic-shape rate' command.<br /><br />How would I enforce a maximum bandwidth on a metro e connection using a Cisco 2811 and 1812?<br /><br />Thanks for your time. 30 naikumarNov 15, 2009, 9:35pm PSTHi,<br /><br />Do you want to regulate your out going traffic at certain rate?.<br /><br />You can configure shaping on 2811/1812 using MQC (Modular QoS command).<br /><br />Below link will guide you to configure shaping using MQC.<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/docs/ios/qos/configuration/guide/reg_pkt_flow_per_cls.html')">http://www.cisco.com/en/US/docs/ios/qos/configuration/guide/reg_pkt_flow_per_cls.html</A><br /><br />HTH,<br />Nagendra<br /> amit.bhagatNov 11, 2009, 4:54am PSTHi All,<br /><br />I noticed a really unusual thing. When multipath is enabled between eBGP and iBGP, the AD of iBGP path also appears as 20 instead of 200. However, if eBGP path is not available and only iBGP path is available, the AD becomes 200.<br /><br />This seems like a bug. Any ideas?<br /><br />Here's the sample output- Notice network 10.2.2.0 in the both outputs.<br /><br />PE1#sh ip route vrf CUST1<br /><br />Gateway of last resort is not set<br /><br /> 172.64.0.0/30 is subnetted, 3 subnets<br />C 172.64.1.0 is directly connected, Serial0/0<br />B 172.64.2.0 is directly connected, 00:22:35, Serial0/1<br />B 172.64.3.0 [200/0] via 2.2.2.2, 00:04:33<br /> 10.0.0.0/24 is subnetted, 2 subnets<br />B 10.2.2.0 [20/0] via 172.64.2.1 (CUST2), 00:22:35<br /> [20/0] via 2.2.2.2, 00:00:27<br />B 10.1.1.0 [20/0] via 172.64.1.1, 00:22:45<br /><br /><br />PE1#sh ip bgp vpnv4 all 10.2.2.0<br />BGP routing table entry for 1:1:10.2.2.0/24, version 24<br />Paths: (2 available, best #2, table CUST1)<br />Multipath: eiBGP<br />Flag: 0x800<br /> Advertised to update-groups:<br /> 1<br /> 65002, imported path from 2:2:10.2.2.0/24<br /> 2.2.2.2 (metric 11) from 2.2.2.2 (2.2.2.2)<br /> Origin IGP, metric 0, localpref 100, valid, internal, multipath<br /> Extended Community: RT:2:2<br /> mpls labels in/out nolabel/20<br /> 65002, imported path from 1:2:10.2.2.0/24<br /> 172.64.2.1 from 172.64.2.1 (10.2.2.1)<br /> Origin IGP, metric 0, localpref 100, valid, external, multipath, best<br /> Extended Community: RT:1:2<br /><br /><br />Regards,<br />Amit. 30 shivluNov 11, 2009, 6:06am PSTHi Amit<br /><br />Please post the diagram.<br /><br />regards<br />shivlu jain giuslarNov 11, 2009, 9:55am PSTHello Amit,<br />I don't think it is a bug.<br /><br />you have asked to perform iBGP and eBGP multipath if the iBGP prefix would be presented with natural AD 200 it wouldn't be installed in routing table and you would see only the eBGP path coming from CE node.<br /><br />it is a way to fool IP routing table mantainer <br /><br />Hope to help<br />Giuseppe<br /> amit.bhagatNov 11, 2009, 2:39pm PST@Giuseppe- I have to agree with you. To support your claim, when I made the eBGP route unavailable, the routing table stored the iBGP path with correct AD i.e. 200.<br /><br />@shivlu- Here's the implementation mate.<br /><A HREF="javascript:newWin('http://sites.google.com/site/amitsciscozone/home/bgp/bgp-multipath-load-sharing-for-both-ebgp-and-ibgp-in-an-mpls-vpn')">http://sites.google.com/site/amitsciscozone/home/bgp/bgp-multipath-load-sharing-for-both-ebgp-and-ibgp-in-an-mpls-vpn</A><br /> shivluNov 12, 2009, 1:28am PSTAmit You are right. I need to simulate the same and will come back with results.<br />Could you post the IOS name.<br /><br />regards<br />shivlu jain amit.bhagatNov 12, 2009, 4:02am PSTHi Shivlu,<br /><br />I tried on c2691-spservicesk9-mz.124-25a.bin.<br /><br />Regards,<br />Amit. sr1405268Nov 8, 2009, 2:05am PSTHi,<br /><br />What considerations should be kept in mind when using BGP as the PE-CE Protocol.<br /><br />I have a simple setup with a Main Site Connecting Regional Offices using MPLS as Primary and DXX (Leased line) as Backup.<br /><br />I think I can modify the AS-Path to make the MPLS Cloud Primary and DXX as Backup<br /><br />Any other thing I need to take care of ?<br /><br />Thanks<br /><br />Zeeshan 30 jcozzupoliNov 9, 2009, 3:07pm PSTHi,<br /><br />There are a few things to consider:<br /><br />- Exit points out either link from within your network using Local Preference.<br /><br />- Entry points for return traffic over either link, best to use MED.<br /><br />- If other sites are also running BGP as their IGP, you'd need to think about using the same ASN or different ASN to them. If you go with same ASN, you'd need to ask the provider to configure "as-override" or you could configure "allowas-in" which forces BGP to allow routes in that contain its own ASN (which is dangerous).<br /><br />- Also, remember convergence times aren't the quickest with BGP, so you may need to tweak timers such as "bgp graceful-restart" and also, simple keepalive and holdtimers.<br /><br />There are just a few things I could think of for now, as I don't know the size and complexity of your network.<br /><br />Cheers,<br />Joe. shivluNov 9, 2009, 9:36pm PSTAre you going to run BGP with backup?<br /><br />regards<br />shivlu jain kelsey.timNov 9, 2009, 1:02pm PSTI have a SUP720(port based mpls) and a SUP2 (SVI based mpls.) I can get the VC to come up if I wanted to do one vlan accross my network.<br /><br />I am having issues getting a port to trunk across my EoMPLS to a 7200 router. <br /><br />Currently, I have a Flex-wan module for the SUP2 side, which supports this, but I am getting different views based off the QinQ.<br /><br />Any ideas? 30 jcozzupoliNov 9, 2009, 2:51pm PSTHi,<br /><br />Have you tried changing the MTU sizes to 9216?<br /><br />Also, what do you mean by "but I am getting different views based off the QinQ."?<br /><br />Joe. kelsey.timNov 9, 2009, 5:41pm PSTNo I kept the MTU at 1500. <br /><br />What I mean by that is that it seems everyone is doing it slightly different from another. <br /><br />I've been all over Cisco's site and I cannot find any clear definite way that other engineers are setting this up.<br /><br />Also, it seems that this topic is considered part of two different departments at Cisco, so not one department is able to go over the entire setup because they either know MPLS or they know dot1q.<br /><br />I just wish there was a document that was like, "this is what you need to do this." news2010aNov 6, 2009, 9:22am PSTFolks, the difference between MPLS frame mode vs MPLS cell mode is that the cell mode is applied in ATM (which operated with fixed size cells)? Then the frame mode term is related to the PDU "frame" which is how data is transmitted at layer 2. Is that all? 30 marikakisNov 6, 2009, 11:25am PSTIn most cases you have frame mode MPLS. In this mode MPLS typically adds its own header information (MPLS shim header) between the Layer 2 header and the Layer 3 header. <br /><br />In ATM networks you can run either frame mode MPLS or cell mode MPLS. In a typical ATM setup you have routers at endpoints and an ATM cloud between the routers. The ATM cloud consists of ATM switches. The difference between the modes has to do with the capability of the intermediate ATM switches to speak MPLS. <br /><br />If the ATM switches don't speak MPLS, you can run frame mode MPLS between the routers at the level say of ATM AAL5 PDUs (instead of cells) and this is transparent to the ATM switches (the switches do not know you run MPLS over them). If the intermediate ATM switches do speak MPLS, then you could run cell mode MPLS with the MPLS label information being mapped to the VPI/VCI field of the ATM cells. giuslarNov 7, 2009, 4:12am PSTHello Marlon,<br /><br />in cell mode MPLS the external label is resolved in the ATM VPI/VCI.<br />from an ATM point of view the Label switch path is considered an SVC switched virtual circuit that is setup on demand (when traffic needs to be sent).<br />the ATM LVC is negotiated over the PVC where label binding happens.<br /><br />so in cell mode we have:<br />a PVC for signalling and setting up dynamic VCs on demand<br />a range of VPI and VCI values that represent the possible vpi#, vci# used for a specific Label SVC.<br /><br />for saving on label space that is small labels/SVCs are assigned on demand when traffic is needed to be sent to intended destination.<br /><br />this is quite different from frame mode where the MPLS label is assigned in an unsolicited way: that is regardless of traffic needs.<br /><br />in frame mode a device can store received labels even if not used (not on current best path) this is called liberal retention.<br /><br />just to give some figures:<br />in frame mode typical label space can be 1000000 of labels<br />in cell mode depending on the settings vpi and vci ranges cannot cover a so big space.<br /><br />Cell mode has probably been historically the first implementation of MPLS but it is becoming rare.<br /><br />signalling overhead is quite big for this on demand setting up SVCs that doesn't fit with topology driven CEF.<br /><br />Hope to help<br />Giuseppe<br />urs.shrestharanaNov 1, 2009, 6:57am PSTA present network scenario, i have a core router (VXR A) with a link connected to MPLS network M-1 Router an e-BGP (instance X), all the customer connected to MPLS network uses instance X of M-1 to VRX A for Internet reach ability. <br /><br />Now changed for high availability network, i will be adding new Core Router VXR B. For this, VXR B will also be connected with MPLS network M-1 router an e-BGP (Instance Y), we have numbers of customer connected to our MPLS network example Customer A and Customer B, my requirement is that both the instance M-1 to VXR A and M-1 to VRX B should be on Active that means Customer A should be going through Instance X and Customer B should be going through Instance Y. In case of link failure between any VXR (A and B) to M-1, route of any customers should be going through the live one instance.<br /><br />Here by I have also attached the network diagram.<br /><br />Hopping for your help for implementing such network scenario.<br /><br /><br /><br /><b>Attachment Keywords : </b> <br />1) VRF_Load_Balance.jpg<br /> VRF_Load_Balance.jpg123006image/pjpegimage3628711/01/2014no30 laaubertNov 1, 2009, 9:08am PSTHi,<br /><br />For each eBGP session, you could set a different BGP community for the default-route so the remote PE could set a different local-pref based on the community. It will allow you to load-balance your customers on both SP and provide redundancy as well.<br /><br />HTH<br /><br />Laurent.urs.shrestharanaNov 1, 2009, 9:59am PSTHi Laurent, <br /><br />Thank you for the response. <br /><br />I am not much good in MPLS VRF and its route policy. As of now our MPLS network is very simple, all the customers are bounded in single Instance ( instance X). I am not getting the exact clue how to accomplish the above scenario. Could you please guide me providing the configuration in sort ( simple example) and the related link.<br /><br />Thank you,<br /><br />Uttam<br /><br />urs.shrestharanaNov 2, 2009, 12:44am PSTHello Experts,<br /><br />Is there any solution if anyone have ? please help.<br /><br />Thank you,<br /><br /> giuslarNov 2, 2009, 2:26am PSTHello Uttam,<br />it is not clear if your customers are in different VRFs or they are in global routing table.<br /><br />router M-1 is the one that can associate different attributes to the default routes as suggested by Laurent.<br />a) global table<br />router M-1<br /><br />router bgp yourAS<br />neigh VXRA.ipaddr remote-as 100<br />neigh VXRB.ipaddr remote-as 200<br /><br />!<br /><br />expanding Laurent's suggestion:<br />neigh VXRA route-map setComVXRA in<br />neigh VXRB route-map setComVXRB in<br /><br />ip prefix-list only-default permit 0.0.0.0/0<br /><br />route-map setComVXRA permit 10<br />match ip address prefix only-default<br />set community yourASN:101<br /><br />route-map setComVXRB permit 10<br />match ip address prefix only-default<br />set community yourASN:102<br /><br /><br />b) customers in VRF and also eBGP sessions<br /><br />the same idea can be used if you put the two eBGP sessions inside a VRF.<br /><br />address-family ipv4 vrf ToVXRA<br />neigh VXRA.ipaddr remote-as 100<br />neigh VXRA.ipaddr route-map setComVXRA in<br /><br /><br />address-family ipv4 vrf ToVXRB<br />neigh VXRB.ipaddr remote-as 200<br />neigh VXRB.ipaddr route-map setComVXRB in<br /><br />you need also to use route targets that are imported in all VRFs (if you are using VRFs)<br /><br />Hope to help<br />Giuseppe<br /> laaubertNov 2, 2009, 6:31am PSTHi,<br /><br />To complete Giuseppe configuration, here is what is needed on M4:<br /><br />route-map VXRA permit 10<br /> match community 1<br /> set local-preference 200<br />!<br />route-map VXRA permit 100<br />!<br />route-map VXRB permit 10<br /> match community 2<br /> set local-preference 200<br />!<br />route-map VXRB permit 100<br />!<br />ip community-list 1 permit ASN:101<br />ip community-list 2 permit ASN:102<br />!<br />ip vrf CustomerA<br /> import-map VXRA<br />!<br />ip vrf CustomerB<br /> import-map VXRB<br />!<br />...<br /><br />HTH<br /><br />Laurent.urs.shrestharanaNov 2, 2009, 9:06am PSTHi Giuseppe/ Laurent <br /><br />We have all our customers are in VRF let's say vrf link_X (a single VRF), now we are adding one more vrf let's say vrf link_Y. And also all the customers are connected with Static VRF (static PE-CE).<br /><br />The main thing is customers with the VRF link_X uses Instance_X primary for internet traffic and Instance_Y for backup and the customers with the VRF link_Y uses Instance_Y for primary path and Instance_X for backup.<br /><br />I have also attached our exact network diagram herewith.<br /><br />For outgoing traffic :<br /><br />As per Giuseppe -<br /><br />whenever M-1 receives default route from Link X (in Instance_X) and from Link Y ( in Instance_Y) it tag the community 65xx:101 and 65xx:102 respectively.<br /><br />As per Laurent -<br /><br />On any other PE (M-x) routers will import the RT of both Instance_X and Instance_Y. <br /><br />If route with the community 65xx:101 than set the Loc-Pref higher (200) otherwise, make it default for the Customers with the VRF link_X.<br /><br />If route with the community 65xx:102 than set the Loc-Pref higher (200) otherwise, make it default for the customers with the vrf link_Y<br /><br />And finally import it on the appropriate VRF table.<br /><br />Am I right ?<br /><br />I will deploy it on our regular maintains day and update you consequently.<br /><br />Thank you for the Help<br /><br />Uttam<br /><br /><br /><b>Attachment Keywords : </b> <br />1) VRF_Redundancy.jpg<br />VRF_Redundancy.jpg123058image/pjpegimage6427611/02/2014no giuslarNov 2, 2009, 11:58pm PSTHello Uttam,<br />the setup we are suggesting should work.<br /><br />As usual with VRFs you need to pay attention to route targets.<br />Customer VRFs need to be able to import the routes injected by VXRA and by VXRB.<br />This means to use appropriate route targets and you need also to take care of return path.<br /><br />Hope to help<br />Giuseppe<br />kaustav.guptaOct 31, 2009, 5:28am PSTHi,<br />If IOS supports route-refresh capability then can i do away with soft-reconfiguration inbound to save memory and cpu utilization in multi-vrf ce routers.<br /><br />Thanks <br /><br />Kas 30 laaubertNov 1, 2009, 9:02am PSTAbsolutely !!<br /><br />soft-reconfiguration inbound is more seen as a troubleshooting tool and should be enabled only to check what you are receiving and then should be removed.<br /><br />HTH<br /><br />Laurent. e.shik_i-tecoApr 3, 2009, 6:15am PSTThe MPLS-Aware NetFlow feature is supported by 7200, 7500 and 12000 platforms only. Will it be supported by 7600 in the nearest future? 30 laaubertApr 3, 2009, 4:46pm PSTHi,<br /><br />It's available since 12.2(33)SRC:<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/docs/ios/netflow/configuration/guide/cfg_mpls_netflow_ps6922_TSD_Products_Configuration_Guide_Chapter.html')">http://www.cisco.com/en/US/docs/ios/netflow/configuration/guide/cfg_mpls_netflow_ps6922_TSD_Products_Configuration_Guide_Chapter.html</A><br /><br />HTH<br /><br />Laurent. e.shik_i-tecoApr 5, 2009, 11:02pm PSTThe command "ip flow-cache mpls label-positions [label-position-1 [label-position-2 [label-position-3]]]" is availiable neither in c7600s72033-advipservicesk9-mz.122-33.SRC.bin nor c7600s3223-advipservicesk9-mz.122-33.SRC3.bin. Although (MPLS)-aware NetFlow is pointed out in Cisco Feature Navigator in both versions.<br /><br /> laaubertApr 6, 2009, 6:13am PSTHi,<br /><br />I went deeper in my research and unfortunately, the documentation is wrong. MPLS aware Netflow is not supported on the 7600 platform.<br /><br />Sorry for that<br /><br />Laurent. e.shik_i-tecoApr 6, 2009, 6:20am PSTThank you very much for your reply! Do you know whether this feature will be introduced in the next version of IOS? laaubertApr 6, 2009, 6:54am PSTNo plan for now.<br /><br />Laurent. aneufOct 30, 2009, 6:00am PSTHas anyone another idea how to get IP-flow information/statistics out of an MPLS-Stream?<br />Mirror-ports/probes might be a solution too.<br /><br />In case of my customer ingress Netflow is not an option.<br /><br />regards<br /><br />AndreaspsmidcnssOct 27, 2009, 9:30am PSTHello Netpros,<br /><br />Here's the sample scenario:<br />- plain MPLS network (MPBGP + OSPF) with VRF VPNs<br />- VRF XXX configured say on PE1,PE2,PE3<br />- PE1 and PE2 are colocated in the same site: there is a a transit vlan to communicate with a couple of firewalls, mapped to VRF XXX; no CEs, just HSRP/VRRP and static routic redistributed in MPBGP<br /><br />How can I force PE3 (remote site) to choose PE1 as best path to reach the Firewalls, for the single VRF XXX ?<br /><br />Hope it's clear, thanks a lot<br /><br />Andrea 30 giuslarOct 28, 2009, 1:04pm PSTHello Andrea,<br />you should be able to apply a route-map in the vpnv4 address-family where:<br />you check the IP BGP next-hop= PE1 loopback<br />and<br />you check route-target = that of VRF XXX<br /><br />and you set local-preference to 200<br /><br />access-list 11 permit host PE1-loop<br />ip <br /><br />route-map rise-preference permit 10<br />match ip next-hop 11<br />match route-target extcommunity 12<br />set local-preference 200<br />! empty final block to accept all other<br />! vpnv4 routes unchanged<br />route-map rise-preference permit 20<br />!<br /><br />router bgp XX<br />address-family vpnv4<br />neigh RRS route-map rise-preference in<br /><br />this setup should work also with a RRS in the middle unless PE2 is preferred by RRS.<br />Use different RDs on PE1 and PE2 to be sure routes of both are propagated in the domain.<br /><br />Hope to help<br />Giuseppe<br />psmidcnssOct 29, 2009, 7:29am PSTHello Giuseppe,<br /><br />Thank your for your suggestion.<br /><br />Yes, there are two RRS in the BGP cloud, and PE1 is not preferred by PE2. We do use different RDs for a single VRF on PE1 PE2 to increase convergence and let the routes propagate in the domain.<br /><br />Now I come to your suggestion:<br /><br />ip extcommunity-list standard TESTEXT permit rt xxx:xxx<br />route-map rise-preference permit 10<br /> match extcommunity TESTEXT<br /> set local-preference 200<br /><br />this works fine (per VRF local preference)<br /><br />On the other hand:<br /><br />(config-route-map)#match ip next-hop 11<br />% "rise-preference" used as BGP inbound route-map, nexthop match not supported<br /><br />I tried a few PEs, but I got the same error when applying the route-map in vpv4 address-family. <br /><br />I will try to investigate a little more.<br /><br />Andrea giuslarOct 29, 2009, 9:39am PSTHello Andrea,<br />your findings are correct not all route-map statements are acceptable in this direction. <br /><br />In this case you need to apply the route-map outbound on PE1, in this way you rise the local-preference at the origin PE.<br /><br />Hope to help<br />Giuseppe<br />psmidcnssOct 30, 2009, 2:39am PSTGiuseppe,<br /><br />indeed it looks more logic to raise the local preference for a certain vrf (rt) on PE1 (as an example) with an outbound route-map, and works fine with the RRSs too in the middle. PE3 receives a higher local preference and uses PE1 as next hop instead of PE2, that sends routing update with standard local preference.<br />The complete config on PE1 is then:<br />router bgp 65000<br /> address-family vpnv4<br /> neighbor RRS1 route-map rise-preference out<br /> neighbor RRS2 route-map rise-preference out<br /><br />ip extcommunity-list standard TESTEXT permit rt xx:xx<br /><br />route-map rise-preference permit 10<br /> match extcommunity TESTEXT<br /> set local-preference 200<br />!<br />route-map rise-preference permit 20<br /><br />Thanks a lot<br /><br />Andrea tckoon@hpOct 27, 2009, 7:28pm PSTHi all,<br />As we knew when turn off ttl propagation, MPLS network ips is not visible if we perform trace route.<br /><br />If MPLS network with dozen of customers(vrf) and I want specific customer not visible the mpls netwok and some customer do have visibility.<br /><br />How can I achive this ??<br /><br />Regards<br /> 30 naikumarOct 29, 2009, 9:28pm PSTHi,<br /><br />MPLS TTL propagation can only be disabld globally and per VRF is not supported.<br /><br />You may have to use access-list on PE to allow/deny ICMP traffic per customer/VRF. <br /><br />HTH,<br />Nagendra patrickg@layer8llc.comOct 28, 2009, 9:01am PSTI want to run EoMPLSoGRE between 2 data centers terminating the GRE and pwire on the ASR. Couple of questions: 1) Are there any special requirements on the ASR to support this? 2) Any caveats with running this in this manner? <br />My first hop is L3 with my provider and hence the need to run EoMPLSoGRE. Thanks for your help 30 laaubertOct 28, 2009, 5:14pm PSTHi,<br /><br />You need IOS XE 2.4.1 minimum to have this feature.<br /><br />HTH<br /><br />Laurent.<br /><br /><br /> patrickg@layer8llc.comOct 29, 2009, 4:39am PSTLaurent,<br />Thanks for the quick response.<br /> mpalisOct 19, 2009, 10:01pm PSTHello<br /><br />I have 3 STM-1 equal cost links between two GSR12000 within my MPLS core. My problem is that load balancing/sharing of traffic passing via this links is done ONLY in one way and not both ways. OSPF is running as an IGP protocol.<br /><br />Can you please advice of how to achieve balancing/sharing of traffic both ways.<br />Thanks 30virvermaOct 20, 2009, 12:55am PSTWhen multiple TE tunnels have the same cost, traffic can be load-balanced across them. Traffic<br />can also be load-balanced between the native IP path and TE tunnels if the cost of the routing is<br />the same. This situation has some restrictions, however; see the earlier section “Cost Calculation<br />of IGP Routes over TE Tunnels.” When you are load balancing over TE tunnels, the load balancing<br />can even be unequal cost load balancing. The load balancing of traffic is weighted proportionally<br />to the bandwidth requirement of the TE tunnels. If you have one tunnel with 80 MB and one with<br />20 MB of reserved bandwidth, the load-balancing ratio is 4:1, or the first tunnel should get four<br />times more traffic than the second tunnel. However, the load-balancing ratio is an approximation,<br />because Cisco Express Forwarding (CEF) has only 16 hash buckets. See Chapter 6, “Cisco<br />Express Forwarding,” for more on this.<br />When an LSR performs the load balancing over one or more IP paths and one or more TE tunnels,<br />it is always equal cost load balancing. This means that every path gets the same amount of traffic.<br />Multiple TE tunnels can be handy when the amount of bandwidth to be reserved between a pair<br />of routers is more than the bandwidth capacity of the links. You can then just create multiple TE<br />tunnels with each a piece of the required bandwidth. huydien3384Oct 27, 2009, 10:50pm PSTDear VIRVERMA, <br /><br />Which algorithms do cisco routers use for load balancing? <br /><br />Thank you<br /> shivluOct 27, 2009, 11:35pm PSTIts inbuilt algo of cisco which uses 4 bits, thats why you cannot use more than 16 links. It always works in even mode.<br /><A HREF="javascript:newWin('http://shivlu.blogspot.com/2008/11/cef-loadbalancingproblem.html')">http://shivlu.blogspot.com/2008/11/cef-loadbalancingproblem.html</A><br /><br />regards<br />shivlu jain fayyaz_sOct 27, 2009, 6:05am PSTbest way to fwd Layer 2 traffic across layer 3 link whihc is b/w 6500 and 4500 (multiple distribution) switches. while l2 clients directly connected to 4500 and 6500, can say non routed vlans earlier while it was connected at Layer 2 trunk, but now l3 Link only b/w these switches (6500 & 4500). 30 collin_clarkOct 27, 2009, 7:41am PSTCheck into QinQ. I know the 6500 supports it. The 4500 may support it in the new Sup, but I'm not sure, so you'll want to look that up in the Feature Navigator.<br /><br /><A HREF="javascript:newWin('http://en.wikipedia.org/wiki/802.1QinQ')">http://en.wikipedia.org/wiki/802.1QinQ</A> giuslarOct 27, 2009, 7:55am PSTHello Fayyaz,<br />the best solution would be to revert to a L2 trunk between C6500 and C4500: just add one vlan that will be the equivalent of the current L3 link.<br /><br />It is th simplest and most efficient solution to extend vlans between two switches.<br /><br />L2TPv3 may be available on some C6500 but it shouldn't be possible with C4500.<br /><br />802.1QinQ could be a way to do this by using only one additional vlan (the external "customer vlan-id") on the link but you still need a L2 trunk to carry two vlans: the outer vlan for QinQ and the vlan for the current L3 service.<br /><br />Hope to help<br />Giuseppe<br />pt-tac@nextiraone.ptOct 22, 2009, 3:19am PST<br />Hi all,<br /><br />Hi have installed a network made of 7609-S with RSP720 and ES-20G line cards with 12.2.33SRB3. We networ is running mpls and we have several EPL configuring on the router to make stress test. On one test we were sending 64b frames in to the EPL and we notice that the router only swiths 20Mpps that represent a 6.XGbits full duplex of bw consume. Then we change the frame to 128byte to see the beavihor and we notice that the router only switch 12,6Mpps and 7.4Gbits full duplex. I expect to have high values around the 20Mpps or 9,XGbits. For frames start at 256bytes the BW consume reach 9,8Gbits and the Mpps start to decrease as normal. The problem is for frame with 128bytes. <br /><br />Can any one point me to same official test where i can compare this results.<br /><br />Many thanks<br /> 30 jpazahanickOct 26, 2009, 2:19pm PSTNothing official, but I have the following results myself for you to compare:<br /><br />64 byte - 7600Mbps 11310054 frames/sec<br />128 byte - 8500.1 7179158 frames/sec<br />256 byte - 9100.0 4121384 frames/sec<br />512 byte - 9500.0 2232152 frames/sec<br />1024 byte - 9700.0 1161404 frames/sec<br />1280 byte - 9800.0 942312 frames/sec<br />1518 byte - 9800.0 796492 frames/sec<br /><br />Thanksagata.czekalskaOct 9, 2009, 12:02am PSTHi Everybody!<br /><br />I decided to set up the thread because I did not find the clear answer in Cisco documentation to following question:<br /><br />Is it a must to configure Ethernet over MPLS in VLAN mode on the subinterfaces on Catalyst 6500 series switch?.<br /><br />I try to configure EoMPLS between MRV 9000 series and Cisco Catalyst 6500 series switches.<br /><br />I will be grateful for the certain answer.<br /><br />Best regards,<br />Agata Czekalska<br />Technical University of Lodz.<br /><br /> 30 lloyd_andrewOct 9, 2009, 7:27pm PSTAgata,<br /><br />VLAN-based Ethernet over MPLS should be configured on subinterfaces. <br /><br />See link below. <br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/docs/switches/lan/catalyst6500/ios/12.2SXF/native/configuration/guide/pfc3mpls.html#wp1252154')">http://www.cisco.com/en/US/docs/switches/lan/catalyst6500/ios/12.2SXF/native/configuration/guide/pfc3mpls.html#wp1252154</A><br /><br />HTH.<br /><br />-Lloydagata.czekalskaOct 22, 2009, 3:04am PSTHi Lloyd and other Folks!!!<br /><br />I configured EoMPLS between MRV 900 series switch and Cisco Catalyst 6500 switch. <br /><br />On the MRV side VC status is active and VC Out interface is known.<br /><br />But on the Catalyst switch VC status is down and output interface is uknown as well as Tunnel label is not ready.<br /><br />VC labels are exchanged between two peers.<br /><br />I configured Catalyst to perform EoMPLS in the following way:<br /><br />!<br />interface GigabitEthernet3/26<br />no ip address<br />switchport<br />switchport trunk encapsulation dot1q<br />switchport trunk allowed vlan 998<br />switchport mode trunk<br /><br />!<br />interface Vlan998<br />no ip address<br />xconnect 172.30.31.3 500 encapsulation mpls<br /><br />Andrew you wrote that VLAN-based Ethernet over MPLS should be configured on subinterfaces.<br /><br />In this cisco document you are writing about is written that EoMPLS is not supported on Layer 3 VLAN interface.<br /><br />And further is written that we must configure VLAN-based EoMPLS on subinterfaces.<br /><br />So I'm still wondering if it is a must as far as you wrote we should.<br /><br />I'm talking all the time about CE-facing interface.<br /><br /><br />So I've tried to configure VC on the subinterface and Cisco rejected the command which enables the subinterface to accept 802.1Q VLAN packets, because the switch is in VTP server mode.<br />Cisco software advise to change VTP mode to Transparent. But this is the VTP server as I wrote.<br /><br />I also got some information that I can not have EoMPLS and local switching at the same time.<br />As I guess now I have such situation. But I would like to confirm this statemenet by You Guys.<br /><br />Maybe that is why VC is down, because on Vlan 998 hardware is EoMPLS and <br />on the GigabitEthernet 3/26 interface hardware is Ethernet.<br /><br />As a CE device I've got D-link. This is test implementation. <br />Maybe there is something wrong on D-link configuration.<br /><br />At last I want to write that between two MRV devices Vlan-based VC works.<br />Also Port-based VC between MRV and Cisco is fine.<br /><br />If You need current topology or any other information I'm waiting.<br /><br />As always I'm grateful and open for every comment, idea, suggestion.<br /><br />Sending greetings,<br />Agata Czekalska<br />Technical University of Lodz jon@giganett.comOct 22, 2009, 3:44am PSTHi<br /><br />I would not expect xconnect on a interface Vlan to work OK. Not supported, need OSM cards or ES cards(for 7600)<br /><br />Hav you tried MUX UNI? Don't know if it is supported on 6500.<br /><br />****************************************<br />This example for the 7600-MUX-UNI Support on LAN Cards feature shows a physical trunk port used as UNI: <br /><br />interface Fastethernet3/1<br />switchport<br />switchport encapsulation dot1q<br />switchport mode trunk<br />switchport trunk allowed VLAN 200-250<br /><br />interface Fastethernet3/1.10<br />encap dot1q 3000<br />xconnect 10.0.0.1 3000 encapsulation mpls<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/docs/routers/7600/ios/12.2SR/configuration/guide/pfc3mpls.html#wp1406020')">http://www.cisco.com/en/US/docs/routers/7600/ios/12.2SR/configuration/guide/pfc3mpls.html#wp1406020</A><br /><br />Jon shivluOct 22, 2009, 4:09am PSTHi Jon<br /><br />what about L2TPv3? Does 7600 lan support it or not?<br /><br />regards<br />shivlu jainagata.czekalskaOct 22, 2009, 4:31am PSTJohn,<br /><br />If You make me 100% sure tha LAN Cards DO NOT SUPPORT EoMPLS in VLAN mode, because you write you do not expect.<br /><br />Are there any other solutions except buying o proper adapter (such SPA)to configure EoMPLS on Catalyst 6500 switch? <br /><br />Best regards,<br />Agata Czekalska<br />Technical University of Lodz giuslarOct 22, 2009, 12:49pm PSTHello Agatha,<br />you should be able to use xconnect on a physical port or on a vlan subinterface of a port<br /><br />see<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/docs/switches/lan/catalyst6500/ios/12.2SXF/native/configuration/guide/pfc3mpls.html#wp1110361')">http://www.cisco.com/en/US/docs/switches/lan/catalyst6500/ios/12.2SXF/native/configuration/guide/pfc3mpls.html#wp1110361</A><br /><br />Configuring PFC3BXL or PFC3B Mode VLAN-Based EoMPLS<br /><br />When configuring PFC3BXL or PFC3B mode VLAN-based EoMPLS, follow these guidelines and restrictions:<br /><br />•The AToM control word is not supported.<br /><br />•Ethernet packets with hardware-level cyclic redundancy check (CRC) errors, framing errors, and runt packets are discarded on input.<br /><br />•>>>>You must configure VLAN-based EoMPLS on subinterfaces. <br /><br />so if you have PFC 3B or better you should be able to do this.<br /><br />Edit:<br />I see there are problems with VTP mode server.<br /><br />I'm afraid you need to make a choice here.<br /><br />Hope to help<br />Giuseppe<br />agata.czekalskaOct 26, 2009, 12:28am PSTHi Giuseppe and other Folks!!!<br /><br />I would like You to ask if MUX-UNI is an IOS feature or maybe it is a hardware?<br /><br />And if Cisco Catalyst 6500 series switch supports it?<br /><br />Best regards,<br />Agata Czekalska<br />Welcome to the Human Network giuslarOct 26, 2009, 9:48am PSTHello Agata,<br /><br />PFC based = hardware based then also a minimum IOS version may be required.<br /><br />However, it should be an efficient implementation.<br /><br />By the way, this is explained in Jon's post.<br /><br />see<br /><br />This example for the 7600-MUX-UNI Support on LAN Cards feature shows a physical trunk port used as UNI: <br /><br />xconnect 10.0.0.1 3000 encapsulation mpls<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/docs/routers/7600/ios/12.2SR/configuration/guide/pfc3mpls.html#wp1406020')">http://www.cisco.com/en/US/docs/routers/7600/ios/12.2SR/configuration/guide/pfc3mpls.html#wp1406020</A> <br /><br /><br />Hope to help<br />Giuseppe<br />agata.czekalskaOct 26, 2009, 10:24am PSTGiuseppe!<br /><br />I understand that PFC is a piece of hardware, but I did not realize that PFC = MUX UNI. I thouht that MUX maybe software feature, but as soon as it is multiplexer it must be a device.<br /><br />I have clarity now. Thank You Guys for being helpul and patient for beginner in Networking World namly me.<br /><br />Anyway, welcome to the Human Network.<br /><br />Best regards,<br />Agata Czekalska<br />Technical University of Lodz Antonio_1Oct 26, 2009, 8:25am PSTHi,<br /><br />Here is the standard MPLS network with vrf INTERNET:<br /><br />PE2--P2--P1--PE1--INTERNET<br /> |<br /> CPE<br />PE1-routes<br />ip route vrf INTERNET 0.0.0.0 0.0.0.0 --> PE1<br />IP route vrf INTERNET 0.0.0.0 0.0.0.0 -- > INTERNET<br />vrf INTERNET 10.10.0.0/16 connected<br />192.168.0.5 is loopback for MP-BGP<br /><br />PE2-routes<br />ip route vrf INTERNET 0.0.0.0 0.0.0.0 --> PE1<br />ip route vrf INTERNET 111.111.111.111 255.255.255.255 10.10.0.10<br /><br />Route on PE2:<br />ip route vrf INTERNET 111.111.111.111 255.255.255.255 10.10.0.10 is misconfigured because nexthop is from network 10.10.0.0/16 which is directly is connected on PE1, and address 111.111.111.111 is not active <br /><br />When I start the ping from CPE to 111.111.111.111 route on PE1 sends packet via P1, and P2 to PE2. PE2 has route that recursively points back to PE1. Ofcourse there is LOOP in routing network. <br />All devices except CPE are Cisco7600<br /><br />Here are the questions:<br />1) With only one continuous PING packet-size 1500Bytes from CPE toward address 111.111.111.111 link utilisation P1-P2 goes up to 150Mbps. And with pings from more CPEs link utilisation rises. Why TTL doesn't prevent increasing of link utilisation?<br />2) CPU on PE2 goes up to 70-80%. Shouln't Cisco 7600 forward packets in hardware? Or maybe recursive lookup is done with CPU?<br /><br />output from PE2:<br />Sh ip route vrf INTERNET 111.111.111.111<br />Routing entry for 111.111.111.111/32<br /> Known via "static", distance 1, metric 0<br /> Redistributing via bgp 65001<br /> Advertised by bgp 65001<br /> Routing Descriptor Blocks:<br /> * 10.10.0.10<br /> Route metric is 0, traffic share count is 1<br /><br />PE2#sh ip route vrf INTERNET 10.10.0.10 <br />Routing entry for 10.10.0.0/16<br /> Known via "bgp 65001", distance 200, metric 0, type internal<br /> Last update from 192.168.0.5 6d17h ago<br /> Routing Descriptor Blocks:<br /> * 192.168.0.5 (Default-IP-Routing-Table), from 192.168.0.5, 6d17h ago<br /> Route metric is 0, traffic share count is 1<br /> AS Hops 0<br /><br />Regards,<br />A 30Antonio_1Oct 26, 2009, 10:12am PSTsorry, I made a mistake.<br />CPE is connected to PE1 like this:<br /><br />PE2--P2--P1--PE1--INTERNET <br /> | <br /> CPE <br /> bandhani2Oct 23, 2009, 10:53pm PSTHi,<br /><br />I want to measure the real performance measure for R4 voice, signaling, sigtran traffic in my network. What would be the ideal parameters reflecting the real traffic, keeping view the CPU load on 2811 SLA routers.<br /><br />I am using currently 1500 byes per 30 sec with 20 msec of packet interval. But it is consuming too much load, and i want to integrate other services too.<br /><br />Any best practice.<br /><br />Thanks DanilZhigOct 23, 2009, 12:53am PSTIs it possible to make a VRF as a NetFlow and SNMP traps source on 12.2SR IOS release? 30 laaubertOct 23, 2009, 12:41pm PSTHi,<br /><br />As you specify the source-interface, just put this interface in a VRF and it should be fine.<br /><br />I made a quick test for SNMP and it's working. I used the snmp-server trap-source command.<br /><br />HTH<br /><br />Laurent.<br /><br /><br /><br /><br /> francisco_1Oct 9, 2009, 4:50am PSTRunning inter-as MP-BGP multicast between PE's using MDT under the VRF without a problem. Problem is GRE dynamic tunnel is built using my loopback interface in the global routing table and not in the VRF. All PIM/MSDP is done on loopback interface within the VRF between PE's so when i block AUTO-RP messages (224.0.1.39/224.0.1.40) between PE's, AUTO-RP messages is still exchange on loopback interface in the global since MDT is using that to build the GRE dynamic tunnel. Is it possible to be selective on which interface MDT uses to build dynamic GRE tunnel between PE's?<br /> <br />Is manual GRE tunnel the way forward?<br />Please let me know if you need clarification<br />Configs.<br /><br /> 30 francisco_1Oct 9, 2009, 8:24am PSTafter researching this i now undertsand that the loopback in the global is used because that's what BGP uses to established adjacency between PE's. so MDT uses that. <br /><br /><br /><br /> laaubertOct 14, 2009, 6:45pm PSTHi Francisco,<br /><br />MDT group is used to encapsulate customer multicast traffic so it's part of the SP domain or the global routing.<br /><br />In case of Inter-AS Option-B, you need to use the RPF-vector PIM extension so you don't have to redistribute the PE loopbacks between both AS.<br /><br />HTH<br /><br />Laurent. danailpetrovOct 22, 2009, 5:39am PSTAs far as I can remember you can do this by choosing a bgp next-hop option located in VRF configuration.<br /><br />for instance:<br /><br />ip vrf a<br /> bgp next-hop Loopback10<br /><br />Hope it helps!<br /><br />Kind regards,<br />Dani hitesh.vinzodaOct 22, 2009, 1:15am PSTHi,<br /><br />I m trying to create a psuedowire between 6509 and 3845 which are connected by means of IPV4 network. 6509 is running with code as 12.33SXI and i found that the encapsulation method under pseudowire class is only mpls not the l2tp.. <br />If i configure the psuedowire class with encapsulation with mpls and network between 6509 and 3845 is not mpls enabled...will it work..?? is there any way to set the encapsulation to l2tp instead of mpls ?<br /><br />if not any workaround ..<br /><br />Regards<br /><br />Hitesh 30 francisco_1Oct 22, 2009, 1:56am PSTyou need mpls label switching to use mpls encapsulation. For cat6500 i dont think the 12.33SXI suuport L2TPV3 VPN. <br /><br />use the Cisco Feature Navigator to search for the feature you need for you switch. This tool is very useful and i encourage people to use it... cypticOct 19, 2009, 6:45am PSTHi! <br /><br />We are setting up xconnects from a physical interface and the EoMPLS TE tunnels are connected over vlan interfaces with trunks in both ends. <br /><br />Do i also have to configure the VLAN mode in someway to get this running? <br /><br />A lot of our tunnels works perfecly but now when we are setting up new tunnels the incoming unicast traffic does not enter from the physical interface to the VC in both ends. <br /><br />When doing packet capture from the incoming interface i can see unicast packets coming in but the the VC packet counters doesn't increase. <br /><br />Multicast traffic are running perfect between the CE's and the cdp neigbour is shown in both ends. <br /><br />Is there any known issue when connecting EoMPLS with xconnect from a physical interface over a trunk with SVI's? <br /> 30 laaubertOct 21, 2009, 5:53pm PSTHi,<br /><br />On 7600, PFC does not support xconnect on SVI interface, you need a SIP card for your core link to support it.<br /><br />HTH<br /><br />Laurent. cypticOct 22, 2009, 12:49am PSTHi!<br /><br />We are not setting up the xconnect on the SVI:s but the OSPF routing runs between them. <br />We have now solved the problem. There was a bug in the software.<br />agata.czekalskaOct 19, 2009, 3:55am PSTHello everybody!<br /><br />How to enable rsvp to start signalling on interfaces. I found in cisco documantation something like that:<br /><br />ip rsvp bandwidth [interface-kbps] [single-flow-kbps] <br /><br />My question concerns necessity of setting bandwidth.<br /><br />Best regards,<br />Agata<br /> 30 giuslarOct 19, 2009, 6:19am PSTHello Agata,<br />see this document<br /><br /><A HREF="javascript:newWin('http://www.cisco.com/en/US/tech/tk436/tk428/technologies_configuration_example09186a0080093fd0.shtml#configfiles')">http://www.cisco.com/en/US/tech/tk436/tk428/technologies_configuration_example09186a0080093fd0.shtml#configfiles</A><br /><br />you need on physical interfaces:<br /><br />mpls ip<br /> mpls traffic-eng tunnels <br />+<br />ip rsvp bandwidth value1 value2<br /><br />also on the routing protocol you need to enable MPLS TE extensions, to define the MPLS TE router-id;<br /><br /><br />first value is the total BW resources on the link, second value is max BW that can be used by a single flow / single tunnel<br /><br />MPLS TE like MPLS LDPs are unidirectional and resources are consumed per direction.<br /><br />Also notice that bandwidth value on an MPLS TE does not imply a policer: that is on an MPLS TE with bw 5 Mbps you can put 20 or more Mbps of traffic.<br />Call Admission Control is performed at tunnel setup.<br />This unless using auto-bandwidth.<br /><br /><br />Hope to help<br />Giuseppe<br /><br />agata.czekalskaOct 19, 2009, 8:37am PSTHi Giuseppe and other Folks!!!<br /><br />The problem is a little bit different.<br /><br />I'm under test implementation of Regular Path Protection (Cisco calls it like taht). Topology consists of one Cisco Catalyst 6500 series switch and four MRV boxes. I have a ring with five nodes.<br /><br />Ring goes like that: Cat6500-MRV1-MRV3-MRV2-MRV4-Cat6500.<br /><br />What I want to do is use MPLS TE feature: path protection. <br /><br />I set up one VC between two MRV switches(MRV3 and MRV 4, so MRV2 is between them). I configured a trunk (which is a TE LSP according to MRV paper). I have also created two rsvp paths, one of wchich plays primary role. The second one acts as a backup path. Both are strict at the moment. <br /><br />I can now transfer some traffic through created VC, but to simulate the backup path I need to anable rsvp protocol on Catalyst 6500 as far as secondary path goes through the Cisco device.<br /><br />Question is: how to make Cisco switch start signalling RSVP messages? <br /><br />As far as I know we create tunnels at the end devices. In my configration Cisco is a transit device. If I issue ip rsvp on interfaces it does not still exchange RSVP messages with other devices.<br /><br />As always gratful for answers, comments, ideas.<br /><br />Best regards,<br />Agata Czekalska<br />Technical University of Lodz paluchpeterOct 20, 2009, 1:46am PSTHello Agata,<br /><br />As far as I know, the RSVP is started on interfaces as soon as MPLS-TE is enabled on them and in the routing protocol. You should not need to use any "ip rsvp bandwidth" command to have the MPLS-TE running. When I teach about the MPLS-TE, I teach the most simple configuration without using any "ip rsvp" commands and it works.<br /><br />Make sure that your MPLS-TE support is correctly turned on exactly as Giuseppe has suggested. Also verify that individual routers are recognized in the "show mpls traffic-eng topology" output.<br /><br />Best regards,<br />Peter<br /> giuslarOct 20, 2009, 11:19pm PSTHello Agata,<br /><br />>> Question is: how to make Cisco switch start signalling RSVP messages? <br /><br />you need to create a tunnel to have RSVP to start its activity.<br /><br />my understanding is that RSVP has no hellos concept.<br /><br />when you create the tunnel messages for setting up the tunnel startup.<br /><br />Peter:<br />ip rsvp bandwidth is still needed to be able to accept a new tunnel setup.<br /><br />Hope to help<br />Giuseppe<br /> paluchpeterOct 21, 2009, 7:12am PSTHello Giuseppe,<br /><br />Regarding the RSVP Hellos - the original specification in RFC 2205 indeed does not specify any hello mechanism. For MPLS-TE, the RFC 3209 specifies an optional Hello mechanism (it is not mandated than any implementation uses it).<br /><br />Regarding the "ip rsvp bandwidth" command: I am completely sure that I do not need to use it <b>if</b> I do not configure the bandwidth requirement of a tunnel. I have verified it right now both with explicit and dynamic path option. Of course, no bandwidth guarantees can then be provided. Certainly, if a tunnel bandwidth requirement is given by the command "tunnel mpls traffic-eng bandwidth", then it is absolutely necessary to use the "ip rsvp bandwidth" command as well, otherwise the reservable and available bandwidth on each MPLS-TE-enabled interface is 0. But if you just need to specify explicit paths without bandwidth requirements than you can happily go without using the "ip rsvp bandwidth" command.<br /><br />Best regards,<br />Peter<br /> giuslarOct 21, 2009, 10:30am PSTHello Peter,<br />I always used a small (1 Mbps) bandwidth setup or auto-bandwidth in my tunnels.<br /><br />MPLS TE CAC is far less smart then ATM CAC because (at least some years ago) it couldn't take in account real traffic volume using the tunnel (even with auto-BW you can provide a max value if I remember correctly).<br /><br />It is funny to discover that missing statement is like 0 BW request.<br /><br />About RSVP activity: I just tried to point out that is different then a standard routing protocol that is you can capture RSVP messages but they relate to configured tunnels if we start with a network that is ready for MPLS TE but with no tunnel configured and we try to capture signalling protocols we shouldn't see RSVP messages.<br /><br />Hope to help<br />Giuseppe<br /><br /><br />agata.czekalskaOct 19, 2009, 2:42am PSTHi everybody!<br /><br />I would like to ask about what reasons do rsvp nodes have to not exchange rsvp packets?<br /><br />Every ivolved switch in mpls domain runs rsvp to extend MPLS network to TE (Traffic Engineering), but none of them sends rsvp packets.<br /><br />If the routing protocol (in this case IS-IS) has to be extended to advertise te information it is extended according to my best knowledge.<br /><br />What I want to finally do is to implement regulat LSP protection, by using strict and loose path. <br /><br />Best regards,<br />Agata Czekalska<br />Technical University of Lodz cisco@sbgnet.comOct 12, 2009, 4:08am PSTWe currently connect our remote sites over an ISP's hosted MPLS network. Occasionally we run into a website not working and all signs point to the MPLS network. I've seen this problem before with IPSEC VPN tunnels but not across the MPLS connections. Has anyone experienced similar issues?<br /> 30 kgodfrey@itcdeltacom.comOct 14, 2009, 11:19am PSTYes, definitely if your provider has links with mtu's say set to 1500. The mpls label is 4 byte's causing websites that set the do not fragment bit to be dropped while traversing those links. This would cause some sites to work and some not.<br /><br />Thankssankararaman.kOct 18, 2009, 9:21pm PSTHi <br /><br />You would experience MTU issue more in case of IPSEC VPN since it carries more overhead compare to MPLS VPN. IPSEC overhead is 46bytes while MPLS is 4bytes.<br /><br />Also requires your ISP to set higher mtu > 1500 size across the path.<br /> <br />Thanks <br /><br />')