Had an overview studying for the 301 exam which I passed Friday. Moving onto 350-401 Encore exam. Creates a mapping between VXLAN and underlay IP addresses.
Thank you very much for this really great explanation. I have a question: what if our LISP domain is very large and is connected via different routers to other various non-LISP domains thru different routers ? This would mean that we essentialy would end with a lot of "defaoult" routes. MS/MR servers should help here I guess.
Man you are amazing...good content..i follow you in CBT for SDWaN..Sorry I missed to join live. I will target next week..I feel that a you got a Jeremy inside...BTW, you sounds like one of my colleagues from PA lol
Hi Kish, thank you for the great explanation. I was just wondering that LISP was created so Core routers could be a bit cheaper, as they do not need full routing table. But would MPLS do the same? Meaning if the Core router will only switch label it didn't have to have the full routing table. So I do not understand what is the advantage of LISP vs MPLS in this case.
This is great material, thank you! Studying for DCID/ DCCOR and there are very few videos out there designed for DCID/ DCCOR. Any ideas of a free platform for more vids on those exams? I know CBT Nuggets is top of the top, but was wondering if there is anything freely accessible you may know about? Cheers :)
@@knight2000-NC I've only seen DMVPN used across WANs, not on campuses. Mostly in this context I was referring to PtP tunnels as well. SDNs make the quick creation and teardown of tunnels a trivial task.
@@KishSquared makes sense..haven't seen many campus networks bur Id preume they are more flat than WAN stuff, other times I've seen DMVPN is trying to replicate Frame Relay type stuff in labbing(not even sure we can still use ODR for such for autoconfig)
Nice Video. Quick question, What routes the PxTR send to non LISP Domain. This router needs to send all customer subnets to the non Lisp domain to make sure there is connectivity outside LISP domain. Maybe the PxTR requests the entire map information to MRs and then convert it to standard route info and send it via EBGP using the local IP as next-hop? Thanks!
Yes that's correct, the PxTR is going to take routes from the LISP domain and advertise them to the non-LISP domain, via BGP or even EIGRP/OSPF. Gotta let the rest of the world know where the LISP routes are!
I use Corel Painter for the chalkboard, but any art program would work fine. I have a Wacom tablet for my hardware, and I record with OBS and/or Streamlabs depending on whether it's a live stream or not. Thanks!
Awesome content. One question: How does Lisp solve the traffic engineering requirements, which made ISPs to dis-aggregate the prefixes and thus resulted in larger routing tables?
LISP doesn't directly influence the global Internet routing table. Instead, it makes it so only certain SP routers need to store the entire table. By leveraging LISP, the edge SP routers will encapsulate traffic such that the core routers can have a much smaller routing table, only focusing on how to get to the other side of the domain. I hope that helps!
@@KishSquared Thanks for your reply. The chalk-board look and feel within the Corel Painter gives a nice old school touch. Very nice!! Can be that I like this, since I am also little bit begin to be older :-) Keep up this nice detailed video's. Like that very much!!
Right! And the duplication of terms and ideas also leads to confusion. Lisp and vxlan are pete and re-pete. I can see now the pieces that led to my confusion in the first place.
Hi, Kish, really good videos. I was wondering if you could help me with this: In Ciscopress' ENCOR guide, LISP routing architecture is explained as follows: "In traditional routing architectures, an endpoint IP address represents the endpoint’s identity and location. If the location of the endpoint changes, its IP address also changes. LISP separates IP addresses into endpoint identifiers (EIDs) and routing locators (RLOCs). This way, endpoints can roam from site to site, and the only thing that changes is their RLOC; the EID remains the same." I don't understand this. Assuming that in this context EID = Endpoint IP Address, how can the EID not change if the client roams from one RLOC to another? How can I client keep its IP if it moves from one subnet to another?
So we need tunnel to every ETR/ITR ? If MS don't have route then the router will sent packet to PETR, but how router get to know about appropriate PXTR??
If a router already has the route does it still reach out to the MS/MR to verify it? If it doesn't I don't fully understand the benefit of LISP. What I am really asking is does it force essentially centralized routing by forcing all routers in the "LISP domain" to verify routes? I understand that it is used to not drop routes if the router doesn't know the route, but I thought the point of it with SD-Access is to be a more efficient form of routing.
What the router would do if there are many PxTRs and it gets a negative reply from MS/MR? Which PxTR he would choose to send a packet? It could be a lot of them, and they could lead to different AS, which leads to suboptimal routing. What's the trick?
They are very different, actually! MPLS reduces multiple VRF/routing domains to a common method of quickly switching via labels. LISP reduces the routing table sizes within a domain and thereby reduces lookup latency. They work great together!
@@KishSquared but if the edge routers need to have the full routing table to register to MS/MR and the Core switch MPLS labels, what is the advantage of having them together?
I have a question .. if the EID was mobile or say roaming ..and the device keep its eid (ipv4) what is the point of tunneling to rloc (encapsulation original packet of eid device) if the rloc will not be to route it to any of its local subnets as the eid ipv4 after roaming doesn't belong to any of these local subnets
LISP reduces routing table size in ISP field by searching R10 ( in our example) not subnet A attached to it. but to reach R10 loopback or /30 prefixes all in-path routers must know the path thus routes. how LISP reduce routing table size?
We need a LISP plug-in living resident within an office suite (like the one in a common CAD program). Then LISP will take off from problem solvers everywhere.
Great stuff. Learned this once when it first came out. And somehow technical trainers tend to try and talk to technically and not juat break it down and expkain in plain english for us visual thinkers. Feel I am finally able to piece all the technology features together how they interconnect.
Hey Jeff, been trying to track you down for a quick bit of guidance regarding one of your UCS courses where i'm sort of stuck. Super lame detail...As you start using USCPE you start in a semi clean slate where it only shows the two FI's etc....The Emulator even the same version you're using out the gate comes with a ton of pre-configured hardware and connections etc. I started to try to guess if you deleted what, or disconnect (learned how) virtual devices to get to what state it was in to then follow your course, but its a mess. Can you sort me out with either how to get it to the slicked state you start your demo's in, and or maybe you can link for me a backup of that config i can import to be in lock step? It would just be so much cleaner to have it the same to follow alone without all that noise and or guess how to set it up similar etc. Let me know man. ;)
Suppose i have 10 vlans on legacy switch and they want to communicate fabric envorinment where VM host are depolyed in one of vlans how this communication works ?
Each legacy VLAN will need to tie into a VXLAN ID. The VXLAN will carry that L2 traffic through the fabric to wherever it needs to go. We can think of VXLAN as a 'glue' that connects legacy VLANs to one another through the L3 domain. HTH!
You mentioned misinformation regarding tunneling and MTU size. VTEPs can not fragment packets, that's explicit in the RFC and we've had issues with applications not working because the MTU wasn't adjusted. Yeah the tunnel comes up fine, but high level applications have issues. Otherwise great video, liked the LISP explanation
@@KishSquared Near as I can tell in my labbing thus far, if the packet is less than the set MTU on the tunnel, but greater than what the physical egress interface can accommodate with the additional headers, the egress interface will fragment even if the df-bit is set in the original ip header, because it's not set in the gre outer ip header. But I might be a little unclear what it all means in the big picture. Unless the df-bit is set, and unless I am misunderstanding, the packet is going to arrive at the destination fragmented in both cases, that is, if the packet is larger than the tunnel mtu the tunnel interface performs the fragmentation, and if it's smaller than the tunnel mtu but too big for the physical interface, the physical interface fragments, but in both cases, it's up to the receiving host to put it back together? and notify the sender to reduce packet size? So if I am drawing the correct conclusion, setting the tunnel MTU correctly is all about respecting the df-bit? Your challenge did lead to some insights not previously considered by me, thank you!!
You talk as if LISP isn't designed to be Internet scalable. LISP is not only designed exactly for that, it adds features like active / active multi-homing, network virtualization and address mobility.
Thank you for this video. It's one of the simplest and clearest explanations of LISP that I've been able to find.
I haven't found any video on UA-cam that best explains LISP than this. Thanks so much
agreed
Kish. You have so much deep understanding regarding how overlay and underlay technology works. I believe this is what SD in about
Such a complex topic summarized and explained in very easy way. Thanks a lot.
I like when you see your notes from time to time .. that what real life is .. we're not machines ... and we shall not hide that ..👍👍👏👏
Man, this is so great. The best explanation of something that’s I think is so difficult but think maybe I can have a go after watching
Very well explained..such a great flow to explain such a complex concept. Thank you.
Kish U are always the best
Hello Jeff, thank you for this amazing vídeos, you really helped me out
gracias por el video , excelente explicación , saludos de Perú
Thank you so much! This video is super clear and very detailed! Cheers!
Had an overview studying for the 301 exam which I passed Friday. Moving onto 350-401 Encore exam. Creates a mapping between VXLAN and underlay IP addresses.
thanks Jeff for your help
Amazing explanations.
Keep it up!
Thank you very much for this really great explanation.
I have a question: what if our LISP domain is very large and is connected via different routers to other various non-LISP domains thru different routers ? This would mean that we essentialy would end with a lot of "defaoult" routes. MS/MR servers should help here I guess.
Amazing explanations! Many Thanks !!
Man you are amazing...good content..i follow you in CBT for SDWaN..Sorry I missed to join live. I will target next week..I feel that a you got a Jeremy inside...BTW, you sounds like one of my colleagues from PA lol
Earlier jeremy was my favourite then i saw Jeffa videos this man is just from another dimension
Thank you! You are awesome!
Thank you! Amazing content.
Amazing how this is like dmvpn with its registering/reply for different mappings. Same for MS which is configured statically like NHS & RP
Wow, this is really cool!
Good tunneling mtu discussion as well.
Amazing content. Thanks Jeff for this. :)
Well done, well explained.
Hi Kish, thank you for the great explanation. I was just wondering that LISP was created so Core routers could be a bit cheaper, as they do not need full routing table. But would MPLS do the same? Meaning if the Core router will only switch label it didn't have to have the full routing table. So I do not understand what is the advantage of LISP vs MPLS in this case.
no the Site to Site(VPN Tunnels) replaced the MPLS old day internet was not fast enough they have to use MPLS
This is great material, thank you! Studying for DCID/ DCCOR and there are very few videos out there designed for DCID/ DCCOR. Any ideas of a free platform for more vids on those exams? I know CBT Nuggets is top of the top, but was wondering if there is anything freely accessible you may know about?
Cheers :)
12:38 DMVPN was the ideal practical use case for tunnels prior to SDN I thought
@@knight2000-NC I've only seen DMVPN used across WANs, not on campuses. Mostly in this context I was referring to PtP tunnels as well. SDNs make the quick creation and teardown of tunnels a trivial task.
@@KishSquared makes sense..haven't seen many campus networks bur Id preume they are more flat than WAN stuff, other times I've seen DMVPN is trying to replicate Frame Relay type stuff in labbing(not even sure we can still use ODR for such for autoconfig)
Good stuff! Subscribed!
Nice Video. Quick question, What routes the PxTR send to non LISP Domain. This router needs to send all customer subnets to the non Lisp domain to make sure there is connectivity outside LISP domain. Maybe the PxTR requests the entire map information to MRs and then convert it to standard route info and send it via EBGP using the local IP as next-hop? Thanks!
Yes that's correct, the PxTR is going to take routes from the LISP domain and advertise them to the non-LISP domain, via BGP or even EIGRP/OSPF. Gotta let the rest of the world know where the LISP routes are!
Google said something around 915k at the time i viewed this video! thanks for the video as well :)
What application and hardware do you use for your drawings?
I use Corel Painter for the chalkboard, but any art program would work fine. I have a Wacom tablet for my hardware, and I record with OBS and/or Streamlabs depending on whether it's a live stream or not. Thanks!
thank you sir !! you are the boss
Awesome content. One question: How does Lisp solve the traffic engineering requirements, which made ISPs to dis-aggregate the prefixes and thus resulted in larger routing tables?
LISP doesn't directly influence the global Internet routing table. Instead, it makes it so only certain SP routers need to store the entire table. By leveraging LISP, the edge SP routers will encapsulate traffic such that the core routers can have a much smaller routing table, only focusing on how to get to the other side of the domain. I hope that helps!
@@KishSquared ❤️❤️❤️
thanks for sharing this more than an intro video; wondering what chalkboard software is that?
Thanks for this topic. What i was wondering, what is the "chalk-board program" called you are using? Looks nice ... :-)
Thanks for the comment! I use Corel Painter for the background, but any art program would suffice. I'm using OBS to record the screen while I write.
@@KishSquared Thanks for your reply. The chalk-board look and feel within the Corel Painter gives a nice old school touch. Very nice!! Can be that I like this, since I am also little bit begin to be older :-) Keep up this nice detailed video's. Like that very much!!
Cool, thanks man
Awesome lecture- Thanks
Thanks Kish
Right! And the duplication of terms and ideas also leads to confusion. Lisp and vxlan are pete and re-pete. I can see now the pieces that led to my confusion in the first place.
Or maybe lisp and vxlan are PETR and re-PETR. lol
Hi, Kish, really good videos. I was wondering if you could help me with this:
In Ciscopress' ENCOR guide, LISP routing architecture is explained as follows:
"In traditional routing architectures, an endpoint IP address represents the endpoint’s identity and location. If the location of the endpoint changes, its IP address also changes. LISP separates IP addresses into endpoint identifiers (EIDs) and routing locators (RLOCs). This way, endpoints can roam from site to site, and the only thing that changes is their RLOC; the EID remains the same."
I don't understand this. Assuming that in this context EID = Endpoint IP Address, how can the EID not change if the client roams from one RLOC to another? How can I client keep its IP if it moves from one subnet to another?
So we need tunnel to every ETR/ITR ?
If MS don't have route then the router will sent packet to PETR, but how router get to know about appropriate PXTR??
thank you!!!
Great video, thank you!
If a router already has the route does it still reach out to the MS/MR to verify it? If it doesn't I don't fully understand the benefit of LISP. What I am really asking is does it force essentially centralized routing by forcing all routers in the "LISP domain" to verify routes?
I understand that it is used to not drop routes if the router doesn't know the route, but I thought the point of it with SD-Access is to be a more efficient form of routing.
Great content, Thanks Jeff
What the router would do if there are many PxTRs and it gets a negative reply from MS/MR?
Which PxTR he would choose to send a packet? It could be a lot of them, and they could lead to different AS, which leads to suboptimal routing. What's the trick?
This is great session man.. its very simple and useful.
How long is the Eid stored in an rloc cache before it's removed?
Good question! Entries are cleared after 24 hours of inactivity.
800,000 routes around there somewhere
yes. close to million, not billion
Which Cisco node does LISP reside?
Hi Jeff
thank you for your great explanation
I see LISP and MPLS looks the same ? what is the referent ?
They are very different, actually! MPLS reduces multiple VRF/routing domains to a common method of quickly switching via labels. LISP reduces the routing table sizes within a domain and thereby reduces lookup latency. They work great together!
@@KishSquared but if the edge routers need to have the full routing table to register to MS/MR and the Core switch MPLS labels, what is the advantage of having them together?
I have a question .. if the EID was mobile or say roaming ..and the device keep its eid (ipv4) what is the point of tunneling to rloc (encapsulation original packet of eid device) if the rloc will not be to route it to any of its local subnets as the eid ipv4 after roaming doesn't belong to any of these local subnets
this one hour content is pretty much equivalent to one week training class, as long as you know the basic concept.
What if the provider has multiple PETR connections to service providers?
LISP reduces routing table size in ISP field by searching R10 ( in our example) not subnet A attached to it. but to reach R10 loopback or /30 prefixes all in-path routers must know the path thus routes. how LISP reduce routing table size?
We need a LISP plug-in living resident within an office suite (like the one in a common CAD program). Then LISP will take off from problem solvers everywhere.
Great stuff. Learned this once when it first came out. And somehow technical trainers tend to try and talk to technically and not juat break it down and expkain in plain english for us visual thinkers. Feel I am finally able to piece all the technology features together how they interconnect.
finally LISP is easy !
Hey Jeff, been trying to track you down for a quick bit of guidance regarding one of your UCS courses where i'm sort of stuck. Super lame detail...As you start using USCPE you start in a semi clean slate where it only shows the two FI's etc....The Emulator even the same version you're using out the gate comes with a ton of pre-configured hardware and connections etc. I started to try to guess if you deleted what, or disconnect (learned how) virtual devices to get to what state it was in to then follow your course, but its a mess. Can you sort me out with either how to get it to the slicked state you start your demo's in, and or maybe you can link for me a backup of that config i can import to be in lock step? It would just be so much cleaner to have it the same to follow alone without all that noise and or guess how to set it up similar etc. Let me know man. ;)
Suppose i have 10 vlans on legacy switch and they want to communicate fabric envorinment where VM host are depolyed in one of vlans how this communication works ?
Each legacy VLAN will need to tie into a VXLAN ID. The VXLAN will carry that L2 traffic through the fabric to wherever it needs to go.
We can think of VXLAN as a 'glue' that connects legacy VLANs to one another through the L3 domain. HTH!
You mentioned misinformation regarding tunneling and MTU size. VTEPs can not fragment packets, that's explicit in the RFC and we've had issues with applications not working because the MTU wasn't adjusted. Yeah the tunnel comes up fine, but high level applications have issues. Otherwise great video, liked the LISP explanation
Thanks for the correction! Guess we should just set the MTU and be done with it, then. Will save everyone headaches that way.
@@KishSquared Near as I can tell in my labbing thus far, if the packet is less than the set MTU on the tunnel, but greater than what the physical egress interface can accommodate with the additional headers, the egress interface will fragment even if the df-bit is set in the original ip header, because it's not set in the gre outer ip header. But I might be a little unclear what it all means in the big picture.
Unless the df-bit is set, and unless I am misunderstanding, the packet is going to arrive at the destination fragmented in both cases, that is, if the packet is larger than the tunnel mtu the tunnel interface performs the fragmentation, and if it's smaller than the tunnel mtu but too big for the physical interface, the physical interface fragments, but in both cases, it's up to the receiving host to put it back together? and notify the sender to reduce packet size?
So if I am drawing the correct conclusion, setting the tunnel MTU correctly is all about respecting the df-bit?
Your challenge did lead to some insights not previously considered by me, thank you!!
Did anyone notice the camera not turning off every 30 min 😀
You talk as if LISP isn't designed to be Internet scalable. LISP is not only designed exactly for that, it adds features like active / active multi-homing, network virtualization and address mobility.