I've been struggling with getting my vrfs to exchange routes for a couple days now. I knew I had to be missing some key part of the configuration. I watched a bunch of tutorials that just talked about exporting and importing route-targets. Your discussion of Route Reflectors and vpnv4 configuration solved my issue. Many thanks
Hey there. Great video. Thanks for taking the time. Just a small clarification, for viewers, if I may. At around 1:16:00, it was mentioned that L2VPN requires VRF, RD and RT on PE routers. To clarify: If you are simply creating point-to-point EoMPLS VCs, there is no requirement for BGP; IGP and LDP is all that is needed in the control plane. If you are using VPLS (layer 2 multi-point-to-multi-point), an RD is only required if you intend to use BGP for VC signalling and end-point auto discovery. Otherwise, if your intention is just to do static VPLS, no RD is necessary because no BGP is necessary.
Rob, at 15:07 you mention that there's no need to configure loopback interfaces on the P routers. However, if I remember correctly, for LDP to work, you must have loopback interfaces configured so that it can use them for forming adjancencies between all routers participating in LDP. So doesn't it become necessary to configure the /32 loopback interfaces on the P routers if LDP is being used for label distribution?
Technically no, you don't. Most deployments have them just because it's easier to implement with loopbacks. The loopbacks on the PE's are used for the LSP to be built end to end. You technically don't need /32 addresses either, if you use the P2P network type and have a /24. All in, there are best practice ways to implement it, and I would follow those. If you wanna get into the weeds like I did, feel free to test it out and learn for yourself, it's what I did.
At around the 1:05:30 point Rob you talk about setting up the CE-PE IGP using OSPF. You explain the process of creating a new OSPF process tied to the customer VRF. It was not quite accurate because OSPF does not support address-family configuration. Instead the command should be *router ospf X vrf Y* Where X is the OSPF process number (2). And Y is the VRF name (vrf Cust_A) in other words to initialize the OSPF process the command should be *router ospf 2 vrf Cust_A*
Rob Riker That's how you decided to take my comment? as opposed to the fact I was trying to help anybody following along on GNS3 or something? I learned so much from the video, but I felt I should point that out in case someone gets *stuck*... Quite a defensive response, Oh well...i'll be more careful next time.
+Roman Hoax Nah, I don't get defensive, not much I can do about it so I make light on the situation. You caught something, so be it. I would have corrected myself if I was doing a demo. It was helpful and that was the goal. To your point of OSPF not supporting AFIs, that is incorrect. OSPFv3 in fact does, at the interface level. OSPFv2 only does with Multicast. router ospfv3 1 ! address-family ipv4 unicast exit-address-family ! address-family ipv6 unicast exit-address-family
Rob Riker The video was very helpful indeed. I'm actually doing a config right now in GNS3 which is why I felt the need to mention that, in case others got stuck or were getting error messages. *Yes indeed I should have said OSPFv2 does not support AFIs.* to be honest, I don't think it takes away from the video at all sir. I watched it from beginning to end and its been entirely helpful. :) I'm actually going for my CCNP, so this is beyond scope, but its fun .... :P
+Roman Hoax I agree, I don't think it takes away either. You pointed it out, I was partially incorrect in the syntax, I added a note to clarify. I only throw stuff out there in case a bypasser reads this and feels the need to clarify. Good luck to you in your studies, feel free to comment away ;P Rob
at 9:35 u say custB is 12.1.1.1/30 and custB(left) is 12.1.1.2/30. But dont u think PE4 has to have 12.1.1.2 ? because that subnet has 2 ip addresses and .2 must be on other side of link (PE4) ?
No. Here's why: L2VPN or Layer 2 VPN operates like a switch, the MPLS VPN Provider is a giant switch. Layer 3 VPN operates like a router. The Provider is a giant router. SInce CustA is leveraging L3VPN there needs to be a /30 between both CEs and PEs on their respective attachment circuits. With L2VPN, that is not the case. There is no VRF configured on the PE to the CE. The provider configures the interface facing the customer to transparently bridge or (Xconnect) traffic that is received in on that interface over the L3 core to the other PE that connects to the other CE device. The CE devices get an IP address, not the PE interfaces. This is where the EoMPLS or AToM feature comes into play. The CEs can form an EIGRP adjacency over those interfaces as if those interfaces were connected to a 3750 switch inside the customers network.
- iv just finished ccna im not that into all that vrf and above stuff yet - can ureccomend good book on l2 l3vpn and/or mpls - why eigrp when there is only 2 routers , u can place static route ? tnx
Congrats on the CCNA! MPLS VPN Is a service, there are multiple components to it, IGP+LDP, VRFs, MP-BGP, PE-CE routing, assuming L3 VPN. L2 isn't quite as involved. However, there is no "book" I would recommend, you really need to go through some training on it to see it all put together. Watching a video of someone that knows what to do and explain the pieces is the best in my opinion. You could use a static route, and you will learn as you progress into CCNP that static routing is not scalable. Dynamic routing is scalable and the reason I mentioned EIGRP to begin with.
Can you elaborate on what you mean by "load sharing"? AS Override is a an ISP configuration that strips the AS number the customer sent out and replaces the ASN with the providers. If 2 CE sites have the same ASN, by default updates between them will be dropped. AS override fixes that.
+Rob Riker thank you Rob for AS overide information. load sharing as its done automatically in Eigrp, if BW of the links are same. But here My question is, how do we configure load sharing in l3vpn mpls cloud?
Ok.... sort of a loaded question, I'll break into 2 parts Part 1 - PE-CE routing - If 2 connections of equal metrics are used for IGP, yes ECMP is possible. With BGP it won't be unless you enable multipathing. Lots of inherited caveats come with that, SoO, backdoor links, route feedback. Part 2 - MPLS forwarding - If there are multiple paths in the MPLS core, IGP is used to determine the LSPs. If you use MPLS TE, you can be more specific. So it is possible to load share from CE to CE if designed for it.
I've been struggling with getting my vrfs to exchange routes for a couple days now. I knew I had to be missing some key part of the configuration. I watched a bunch of tutorials that just talked about exporting and importing route-targets. Your discussion of Route Reflectors and vpnv4 configuration solved my issue. Many thanks
Glad it helped!
Very nicely explained, Just like Tom Cruise Mission Impossible. Please keep up the good work.
Hey there. Great video. Thanks for taking the time. Just a small clarification, for viewers, if I may. At around 1:16:00, it was mentioned that L2VPN requires VRF, RD and RT on PE routers. To clarify: If you are simply creating point-to-point EoMPLS VCs, there is no requirement for BGP; IGP and LDP is all that is needed in the control plane. If you are using VPLS (layer 2 multi-point-to-multi-point), an RD is only required if you intend to use BGP for VC signalling and end-point auto discovery. Otherwise, if your intention is just to do static VPLS, no RD is necessary because no BGP is necessary.
indeed and good call out
This video really helped me to understand what's going on inside the provider network. Thanks Rob!
This is decent stuff for l3vpn and mpls understanding..good job Rob
Excellent video, I really enjoy long form vids like these that piece together the bigger picture.
10:41 igp and ldp
27:44 vrfs route distinguishers route targets
Literally this all clicked when you said MPLS is not a protocol, it’s a service.
Excellent video, very good instructor. Learned a lot. Thank you.
Rob, at 15:07 you mention that there's no need to configure loopback interfaces on the P routers. However, if I remember correctly, for LDP to work, you must have loopback interfaces configured so that it can use them for forming adjancencies between all routers participating in LDP. So doesn't it become necessary to configure the /32 loopback interfaces on the P routers if LDP is being used for label distribution?
Technically no, you don't. Most deployments have them just because it's easier to implement with loopbacks. The loopbacks on the PE's are used for the LSP to be built end to end. You technically don't need /32 addresses either, if you use the P2P network type and have a /24. All in, there are best practice ways to implement it, and I would follow those. If you wanna get into the weeds like I did, feel free to test it out and learn for yourself, it's what I did.
At around the 1:05:30 point Rob you talk about setting up the CE-PE IGP using OSPF. You explain the process of creating a new OSPF process tied to the customer VRF. It was not quite accurate because OSPF does not support address-family configuration. Instead the command should be *router ospf X vrf Y* Where X is the OSPF process number (2). And Y is the VRF name (vrf Cust_A) in other words to initialize the OSPF process the command should be *router ospf 2 vrf Cust_A*
+Roman Hoax Someone is always a critic.
Rob Riker
That's how you decided to take my comment? as opposed to the fact I was trying to help anybody following along on GNS3 or something?
I learned so much from the video, but I felt I should point that out in case someone gets *stuck*...
Quite a defensive response, Oh well...i'll be more careful next time.
+Roman Hoax Nah, I don't get defensive, not much I can do about it so I make light on the situation. You caught something, so be it. I would have corrected myself if I was doing a demo. It was helpful and that was the goal. To your point of OSPF not supporting AFIs, that is incorrect. OSPFv3 in fact does, at the interface level. OSPFv2 only does with Multicast.
router ospfv3 1
!
address-family ipv4 unicast
exit-address-family
!
address-family ipv6 unicast
exit-address-family
Rob Riker
The video was very helpful indeed. I'm actually doing a config right now in GNS3 which is why I felt the need to mention that, in case others got stuck or were getting error messages.
*Yes indeed I should have said OSPFv2 does not support AFIs.*
to be honest, I don't think it takes away from the video at all sir. I watched it from beginning to end and its been entirely helpful. :)
I'm actually going for my CCNP, so this is beyond scope, but its fun .... :P
+Roman Hoax I agree, I don't think it takes away either. You pointed it out, I was partially incorrect in the syntax, I added a note to clarify. I only throw stuff out there in case a bypasser reads this and feels the need to clarify. Good luck to you in your studies, feel free to comment away ;P
Rob
Amazing video! thank you! Subscribed!
Other than that, great video :D
Cool thanks
Hey Rob, do you have any videos on IP SLA?
Yes I do, Policy Based Routing covers it. I also have covered it in some of my R&S courses on Rikers Island Training.
Rob Riker can you please leave me the link, because I would like to,watch it.
at 9:35 u say custB is 12.1.1.1/30 and custB(left) is 12.1.1.2/30. But dont u think PE4 has to have 12.1.1.2 ? because that subnet has 2 ip addresses and .2 must be on other side of link (PE4) ?
No. Here's why: L2VPN or Layer 2 VPN operates like a switch, the MPLS VPN Provider is a giant switch. Layer 3 VPN operates like a router. The Provider is a giant router. SInce CustA is leveraging L3VPN there needs to be a /30 between both CEs and PEs on their respective attachment circuits. With L2VPN, that is not the case. There is no VRF configured on the PE to the CE. The provider configures the interface facing the customer to transparently bridge or (Xconnect) traffic that is received in on that interface over the L3 core to the other PE that connects to the other CE device. The CE devices get an IP address, not the PE interfaces. This is where the EoMPLS or AToM feature comes into play. The CEs can form an EIGRP adjacency over those interfaces as if those interfaces were connected to a 3750 switch inside the customers network.
- iv just finished ccna im not that into all that vrf and above stuff yet
- can ureccomend good book on l2 l3vpn and/or mpls
- why eigrp when there is only 2 routers , u can place static route ?
tnx
Congrats on the CCNA!
MPLS VPN Is a service, there are multiple components to it, IGP+LDP, VRFs, MP-BGP, PE-CE routing, assuming L3 VPN. L2 isn't quite as involved. However, there is no "book" I would recommend, you really need to go through some training on it to see it all put together. Watching a video of someone that knows what to do and explain the pieces is the best in my opinion.
You could use a static route, and you will learn as you progress into CCNP that static routing is not scalable. Dynamic routing is scalable and the reason I mentioned EIGRP to begin with.
Hi Rob, Video was very helpful. can u plz explain
1) how to config load sharing on mpls cloud?
2) what is AS overide?
Can you elaborate on what you mean by "load sharing"?
AS Override is a an ISP configuration that strips the AS number the customer sent out and replaces the ASN with the providers. If 2 CE sites have the same ASN, by default updates between them will be dropped. AS override fixes that.
+Rob Riker thank you Rob for AS overide information.
load sharing as its done automatically in Eigrp, if BW of the links are same.
But here My question is, how do we configure load sharing in l3vpn mpls cloud?
Ok.... sort of a loaded question, I'll break into 2 parts
Part 1 - PE-CE routing - If 2 connections of equal metrics are used for IGP, yes ECMP is possible. With BGP it won't be unless you enable multipathing. Lots of inherited caveats come with that, SoO, backdoor links, route feedback.
Part 2 - MPLS forwarding - If there are multiple paths in the MPLS core, IGP is used to determine the LSPs. If you use MPLS TE, you can be more specific. So it is possible to load share from CE to CE if designed for it.
+Rob Riker Thank u Rob. can you please share the configuration for the part 2 scenario?
Good and Useful Video, Thank you.
man where ever i go ,i see u there 😂😂😂😂
How can this be accomplished when you have to traverse the Internet?
If your leveraging MPLS, you won't be going over the internet.
good stuff
Holy cow man. I was going to subscribe and hopefully get you some money somehow but with annoying adds every 10 minutes, I lose focus.
Scratch that... make that every 5-6 minutes.
I will subscribe to INE to get more of your content and not have to deal with ads. Thanks!