RS0156 SD WAN Multicast

RS0156   SD WAN Multicast



welcome to lamb is calm in a lab video series let's disco SD win this is meta your instructor for this video series for a complete list of sd1 videos you can visit a website under the routing switching section there you can also sign up for our newsletters to receive the latest video updates in this video we are going to add yet another network service on top of our Sdn overlay and that is the multicast routing let's take a look at our lab topology here I have the majority of a diagram covered up because as far as this labs concerned though it's irrelevant we still have how Sdn controllers Travie manage people and be smart in the simulated cloud and these are the corresponding IP right here 1 1 0 11 12 and 13 the two sites they will be using in this video is our site ID 102 and 103 with the router BR 2 being the beach cloud and the router bi 3 being the BH 1000 and they're both are part of the SDU an overlay across to transport Internet and MPLS so the two sites have direct IPSec tunnel which is currently able to pass traffic sit in the back we have two windows test machine for site 102 is Windows 10 test 2 at the IP of once a to 17 to 10 and Sai 1 to 3 we got Windows 7 test 1 at the IP of one say 217 310 and those are the machine that we're going to use to source and receive the amount of cast traffic if you've been following our lab video series you might notice one change that we have made in this lab and that is the VLAN ID of the site 103 before I had them all use VLAN 10 just because we trying to consistently apply the interface template to use a sub interface encapsulation VLAN 10 all right with the DA 10 but it's just a way that have the lab setup is that they're all share the same physical switch switch to when they have them VLAN 10 they're actually literally on the same VLAN it wasn't matter because before was using unicast IP all right even though they're in the same VLAN it's to need to hit the right gateway and get out so they weren't really in the issue but now we are dealing with Maori casts if I leave them in the same VLAN 10 then when I sent the malleus traffic from one machine the other one would just received it right away and essentially bypass the entire man so what I had to do just for this lap is that I have to create a separate VLAN wheel and 20 in this case reconfigure the inside interface of the beyond 3 to the SAP interface top 20 and now the two machine cannot see each other as first amount cast trap is concerned and the only way to send the traffic in between is to go through the sd1 itself so I just want to point that out so as far as the Matic as routing is concerned with the Cisco SD LAN it is in the context from the service VPN perspective in our case here we have a service VPN number 10 across all the sites so the traffic has to be either originated and being received on those PPN through the overlay anything that you have comes to know I learned about multicast routing whether there are the PIM protocol and IG MPs to apply with some twist to it especially how the multicast information is exchanged over the SD win first of all there are three most important restrictions that you might want to remember and they are first of all the when each device cannot act as a rendezvous point so you need to have another device to be your RP the second restriction is that only the sparse mode and the source specific modes are supported and in the third is that as our version 18.3 and 16.9 all right first is XD the iOS XE does not appear to support mounted casts and I'll show you in the VLAN interface in a second and that's also a reason why I have to pick the two site that we have the VH because the other sites are all eyes off okay right and they wouldn't be capable of doing now to cast over at the SD win before we go over the operation of the multicast routing let me introduce you to some new terminology and this is specific to the body casts right the first terminology is the replicate do you have a second diagram yes I do so maybe I can use this to explain this terminology first terminology is a replicator replicator concept is unique to Sisko EST un because we're not really running Mattie Cass in the underlay which is to transport the multicast traffic ste needs to be encapsulated inside unicast packet and the job of the replicator is to replicate the amount of cast traffic to the way H that receivers are located alright so the job the replica is basically simulate amout of cast type of behavior from using unicast packet this is very similar to the concept of head and replication and we ran into that I guess in another video Syria was SDA alright it's hard to find access because what happens the same thing the overlay in that case we're dealing with the overlay in the land segment or is not capable of running multicast in the underlay so the head replication concept is that the ingress router also sent multiple packets of what should be a multicast packet to all of the receivers but in this case instead of using the ingress router we have a dedicated router to do that all right so the ingress or the originating router would not do the replication but instead it will first forward the traffic to a replicator for distribution the advantage of doing it this way is you're not locking down the LAN edge because that's what's going to be acting as the ingress router bypassing that responsibility to possibly a more powerful and centralized device alright so that's a replicator the next terminology is the mounted cast service route multicast service route is used for a couple of things first it is used by the LAN edge to inform the be smart on what it knows about the local mounted cast setup this is known as Auto discovery routes this information could be is own PIM status for there and that has pin minimal and run/walk interface or the local rendezvous point if it happens to be located at the site where the rendezvous point is and if any then that way the rest of the network is aware of the router participation in Mountie Cass and the second use of the multicast service route is for the win edge to join the Mountie cast tree by announcing the presence of the vocal receiver okay so does do you think make itself known to the V smart that wants to participate Matti Cass and the other one is advertising a local receiver so it can basically ask for replicated to say in traffic to it now let's kind of go through the process using this diagram right here in this lab set up behalf I'll switch to over here acting as a rendezvous point and the IP address for that is one say 217 2.2 so you can see and it's being tapped by the yellow circle right here marked RP the B are 2 pi/3 will participate in multicast routing br-2 were you going to designate that as now replicator ok replicator is kind of mandatory so you need to assign a router to do that job so the process goes like this first you enable the multicast the pin and the IGMP and the one edge and it's corresponding interfaces in this case now interface is marked as the red dot all right so we have PIM and IGMP enable if there is a local paper it would bring up adjacency in this case like a br 2 it has a neighbor of switch to in this case I'll switch to also act as a rendezvous point so the BR tube will learn about the rendezvous point from the auto RP then through an edge with Santi Auto discovery routes to the V smart in order to join the multicast overlay and also to inform the V smart about the rendezvous point and it's replicator status so in this case the BR 2 will tell the V smart that itself is a replicator and it has an RP of the site right BR 3 just telling the V smart that it wants to be part of them out of cash routing V smart then will relay this information that learns to all the way edge at this point event is just aware of the location of the RP and the replicator and any other wet edge that participate Mountie Cass because in this case br3 and B are two would know that it's basically each other's multicast neighbor and bi3 knows that b r2 is a replicator as well as there's a run every point located at that site that basically sets up the infrastructure of the Mattie caste and your Mattie Cass source one comes along and registered with the RP just like the traditional Mountie caste nothing new there and then the receiver pops up sending IGMP join to the local gateway that has IGMP enable costing the local when edge in this case is beyond three to originated a multicast service route to be smart be smart then for that routes towards the RPE obviously in order to join the Chantry if that's the first time or you can also forward that directly to the multicast source if this is trying to switch to the shortest path tree from the share tree once the tree is joined in the multicast traffic is sent to the source local LAN edge who would then forward to the replicator ok if there's a replicator somewhere else then the br-2 was sent to that replicator and will come back down to be out 3 min this case the br-2 is can a dual-purpose also acting as a replicator the replicator was saying directly to the join router and finally the wine is local to the receiver for with the multicast to the receiver so hopefully the process right there kind of payload a picture of what's going on and you can see that anything within the service site remains the same whether it's the pim adjacency the rendezvous point or the IGMP join or report all right what's different is what's going on on the overlay and the way that the message is being exchanged between the V smart and the win edge now that you understand the concept let's get into the configuration and bring up web interface to the V manage all the multicast configuration is going to be in the form of featured templates and there's actually several of them that we need to configure it the first one is going to be a multi as template and I just want to point now if you select any one of the iOS XE device there is pretty much nothing in here that's really into the multicast or IGMP or PIM template okay and that's why we said that it's pretty much not supported at this time but if you select we have here as B at 1000 or a VH cloud you can see we've got naughty casts IGMP and pin we're going to start with Madi cast okay we're gonna call it LM m casts pH and just FYI the multicast template in the sd1 sense is not really enabling Matic as reality but to enable the replicator function okay and that should be on the threshold is the number of groups the replicator is allowed to accept that multicast groups particularly you can have more than one replicator if you want to perform some kind of load balancing and load sharing if you deal with a large deployment of multicast and that's what threshold is for so we said threshold you can almost guarantee that you would not be overloading the replicator you leave out the default and click Save the next template we're going to configure its PIM select PIM PIM concepts the same as the classic router a Cisco router will call it Alvin p.m. pH you still enable pain based on the interface and then you have an ability to make a case adjustment to the parameters or settings the shutdown will be no this is a global configuration auto RP you probably want to turn that on because is how we are discovering the rendezvous point for the SPT threshold this determined the traffic amount in kilo bits per second before switching to the shortest path tree the default zero means that it will never switch to the shortest path tree it will leave that to default the replicator option in the case that you have multiple replicators this is how you want the amount of cash grouped assigned to them you have an option of let's see this liquid global random and sticky okay granite means that is do a random load bouncing between the multiple replicator sticky means that it's gonna keep using one replicator until it exceeds the threshold before moving on to the next one and then we put it back to the default if the device you're configuring is exclusively used as a replicator only and not really participating the multicast routing these are pretty much to all the config that you need right enable with the Maui cast template and then the PIM globally but here we still need to enable pim on the lan interface of these two routers so we're going to move on to the next section which is the interface we need to provide the interface name because the name has changed for us here we use the dot tan on br-2 and a dot twenty on b3 we can't really make it global so has to be device specific okay PIM interface name I'll just add number one then you guys just down low in the bowl or join put it in the bowl if you want let's go ahead and mark this as optional as well okay let me do it again change my name back great click Add and safe and the last feature that we need to create here is for IGMP and a thousand and the Clow IGMP we'll call it LM IGMP vh shutdown is now enable interface same deal make it device specific if you want to have the interface join to a particular multicast address you can do at drawing group address I guess it's very similar to of the static joys of cisco router let's go ahead and mark that's optional click Add and then save you can see there's a whole lot that you can't get to figure in terms of options they're just basically enable/disable for IGMP and even pim there wasn't a whole lot that you can change now we have to apply the templates to the device starting with the VR 2 which is VH Clow this is going to be applied to the service VPN so we need to find a source VPN right here at IGMP this guy's a replicator 2 so Mountie calves and pain alright GMP mount casts and pin just like that click update we need to provide the interface name for Pam it is yep GE 0 to 10 same here for IGMP so we're now enabling p.m. on the LAN interface and IGMP okay next let's see what it looks like with a config def so the command goes under VPN 10 it's a nada cast replicator and local PIM is all to RP enable interface IGMP enable interface make sure that works and then we're going to go back and repeat on PR three except that the three doesn't need multicast just IGMP and pin because it's not a replicator this one is dot 20 okay and 20 just a quick look yep router Pam there's no replicator command here interface interface alright now we can verify and see what it looks like on the router I bring up the console to I should be our – I don't have console because it's virtual I'm gonna ssh to it okay show him interface yep you can see that the interface gig 0 to 10 right with the IP of 17 to 1 is enable it's got one neighbor which is expected because that's where the rendezvous point is with the switch – with pin the naval dr address that's the IP of the rendezvous point if you show p.m. neighbor off the VLAN 10 off the interface the neighbour address 2 to a 2.2 is the IP right there priority whole time looks good and then let's see if this guy actually learned so we got interface got neighbor the RP mapping from our rendezvous point yep and because we have auto RP enable you can see they've learned from the group 2 to 4 / 4 the RP is the switch – you can do a show multicast which gives you a lot of information over here reverse path forwarding you're pointing towards rendezvous point this resolved next stop count there's only one off the LAN interface which makes sense let me give it oh these because this is the rendezvous point then the reverse path I should point that way alright we should do multipass topology it's very similar to the EM route on the Cisco router because we do auto RP we're seeing the 139 and 140 group for our RP let's join it looks like there's another group that is there I figure out what this one's for note ii/3 9 to 5 5 to 8 5 to 50 bytes there and then this is the information that has gathered itself is the replicator status is up and because we also have the BR 3 participating malic acid knows that's another neighbor the tunnel to that is up so that tunnel to be R 3 + bi 3 is not a replicator let's take a look at the V smart right there I'll use real-time and then look for I think it's a MP Malik as receive autodiscover so it has received the other discover remember the message that the router inform about it schematic a status and information about replicator and local RPE alright let's receive all that from both B R 2 and V R 3 ok so that's good same thing we can lock into the bf3 and do show p.m. interface it has pin minimal on the interface but there's no neighbor ok because you have no PM device over here let's see if this guy has learned about the RP mapping and clearly it has all right it does know about the rendezvous point address 2.2 can you show Maddie Cass Sam deal the reverse path forward is resolved although it has to next stop and this is because between B 3 and B are two there are two tunnels right let me again draw this better the first tunnel is over the mpls and the second tunnel is over the internet which is saying so right here MPLS internet it has joined the other RP it knows about the replicator status is up and it knows about the neighbor who is also being a replicator which is BR to mr. show MP there's a multicast auto-discover alright so this one is local and this one is learned from the be smart the multicast pier so we pretty much have the infrastructure set up the next thing we do is to introduce the medic a source and receiver and do some testing we're gonna use a win 10 test 2 at site 102 s now source so let me bring it up there you go branch number 2 if I make sure that I can ping 3 to n so the unicast routing is in place as you can see that I can ping from 210 to 310 here we're going to use a a very simple mounting as testing application from single wire you can make the the computer be either a server or client right server being the source which will transmit the packet or the client that will be a receiver this guy I'm going to make it a source so select number one by default we use the medic ass IP of 239 0 1 2 with the port of 24 80 which is good enough you can customize it but we don't need to so now we have introduced the source hopefully it has registered with the rendezvous points around every point knows about it and we can look at that I guess we never looked at device I guess a either another console here so I'm gonna take this guy and then click that switch to alright you can look at the PIM neighbor alright its neighboring 2.1 which is br-2 router and then I can do show IP and route 2 3 9 0 1 2 ok so we have the share tree right here and here's the source 210 it's a Windows 10 test to machine incoming on VLAN 10 so now the rendezvous point notes about the source it's just waiting for the receiver to join next we are going to start the receiver and this is going to be on test one over here receiver same application this one's going to be the client it's for number two hit enter okay it's not getting the packets but let's take a look anyway so I should be receiving the or asking for the logic as traffic that match that just the same IP and port that we are sending right now let's take a look at beyond three and see what that looks like first to start with the show I GMT it's the basic mode cast troubleshooting making sure that receiver has make the IGMP request yep you can see that Chimpy is enable on the interface 2.20 but doesn't have a group yep you can see it has a group okay members press in which is au win seven tests one over here so that groups are good as well then let's take a look at OMP multicast route one thing at a time and so it seems like this guy has already sent the multicast service route to the v smart right with the intent to tell it that if there's a local receiver you can see that's the IP of the group into the towards the rendezvous point of 2.2 by the way this is the originating router 0 103 destination which is the reverse path forwarding next top to the rendezvous point and that's the group never look at the V smart and to the OMP multicast receive routes yep you can see there you go there's a column header there so it makes a little bit easier if the originator destination so be smart know where it's to forward that to which group to join and if source not specified at me this is going to join the share tree so let's receive we can also look at the advertised pretty much same type of information towards the rendezvous point that route goes to the peer of 102 and now finally track it all the way to the VR 2 as you can see we're doing like step by step troubleshooting here on paraffin let's see that it receives the multicast service routes by show NP multicast service route and it does you can see it receives it it knows about the receiver at 103 if you do show multicast again we have a share tree over here for the 2 3 9 0 1 2 group with the yep output interface being the BR 3 so knows that if it receives traffic for the group then it has to forward it to BR 3 and the last device for tracing this 2 is the switch to you can see the difference before the chantry has no outgoing interface this time it does ok so right now switch to being around every point is sending traffic out of let me bring up the diagram listen Chantry send it to the VLAN 10 it should hit BR to interface BR to knows that the output interface is BR 3 so it should have send it to be a 3 now it's up to the bi 3 to send it to receiver and what I have done to confirm that the multicast traffic in fact pass through the switched to Windows machine I believe I have the NetFlow enable let me see if I do or not yep so you can see right here take a bit 1 0 5 connects to our ESXi server where the windows 7 is hosted I do have the IP flow monitor to a minor called monitor on the output which means that if I do show flow minor guess monitor cache format table and then look for a 2 39.0 address yep you can see right here that's a flow record of a flow coming from 17 to 10 sending two two three nine zero one two and the packet count keeps going up as well but for some reason is not being detected or received properly by Windows machine here I believe I ran into this issue and it was prepping for the lab spend some time so I was shooting it I think it comes down to either the application or the NIC or the way the ESXi is behaving on this guy but I remember that I left it running for a while I was certain it just started working but it seems like it goes back to not working right now but either way just based on the flow record that we're seeing it appears as if that the multicast is being forwarded to that VLAN 20 is just not being received properly by the Machine either way that's what the output would look like when you look at components whether it's the wind edge the rendezvous point of V smart I guess this is a good step-by-step troubleshooting for you as well they just come down to the end point right now and I'm not going to bother troubleshooting it because I think we verify enough that it's probably the end point issue and by the way I also want to point out that because we are still on the chantry here we are not switching to the shortest path tree okay otherwise the Cline worth trying to join this tree right here at directly although it might not show up though it's just the way we have this set up is on the same subnet so as you can see there isn't a lot of config that you need to do or even can do to get this to work assuming your overlay is already set up and working I did ran into some issues while prepping for the lab and I want to make you aware of first it appears that the when edge can only support IGMP b2 on the interface and as we saw in the IG MP template there wasn't a place where you can switch to v3 when it did the show command on IGMP does it say v2 or anything but since yep right there summary version – so if you have a multicast receiver they used version 3 like most of the Windows machine Windows 7 Windows 10 do they might not be able to communicate okay so what I have to do to even get this guy to join and shape you properly in this lab was changing the system registry and force it to use IGMP b2 and the other issue ran into is that I couldn't seems to get the setup working when Nadi cast source is located at a different site from the RP so for example if we make the Windows 7 over here as a source we wouldn't see the Wynette registered with the RP as we normally expect it hence the receiver would not be able to find the source and receive the Medicus traffic if that happens to not be supported then that might be a certificate limitation and now they're set up that's pretty much a wrap for video st way and multicast you can visit our website to view an extensive list of our lab video so sign up to get access to additional map contents thank you for watching Labatt's calm and I'll see you guys in the next video

Leave a Reply

Your email address will not be published. Required fields are marked *