Great job! I'm a Chinese and I think this is the best video series about vxlan on youtube. I watched other people's videos but I wasn't able to catch up with them...Your illustration and English subtitle are helpful. I have applied for a membership of your website!
When you can teach and explain a subject easy as you do it, means that you really dominate this subject! Great and easy to understand explanation. Thank you
Thank you so very much for this VxLAN series. It is the best I've come across after searching and watching a couple. Thank you for all the other ENCOR topics you have taken the time to present. You're the best
I had this itch to understand VxLan.I am very young in my networking journey. People advise me to stick to routing and switching and then move to security then datacenter, but I try to learn all together. I went through the first 4 videos and things were very clear, for the 4th video I got out my notebook and took notes, now I will do the same for part 6. Thank you for being a part of the learning process. looking forward to more videos on more topics.
I have been working the CCNP Datacenter (350-601), and this video parts where very helpful for me, rather than the official Cisco Cert Guide . Thank you very much !!!
Thanks to comments from Khan in Part 6 of this series, I would like to point out that static ingress replication is now an option as an alternative to multicast. It requires that each peer be manually specified in each VTEP, so it doesn't scale well. My preference is still to use multicast Details here: goo.gl/7fdXgR
Read More at: networkdirection.net/VxLAN+Bridging+Configuration The book (affiliate): click.linksynergy.com/link?id=RL4E*8CmbSY&offerid=145238.2463561&type=2&murl=http%3A%2F%2Fwww.ciscopress.com%2Ftitle%2F9781587144677
How are the 192.168.0.x hosts able to ping each other and how is the same network being used in parallel? Are these subnets being advertised into OSFP with different VRFs?
Hi thanks for your video. It’s really amazing but I have one small doubt in the video. Why do we need any cast getway for this particular data plan learning because you’re not using it right , could you please , is really needed any cast gateway since it’s all learning through multicast
Ok, I must be confused or missing something here, but where is the SVI for vlan 1000? I am assuming vlan 1000 is the 192.168.0.0/24 subnet? I didnt see it in your configurations.
Thanks for Explaining in a detailed manner, Can you please clarify one thing. We cant connect leaf switches directly to each other ri8, then how come you have connected Eth1/49 on Both sides?
Hi, I have tried this lab in EVE-NG Pro seeing your process. I used NXOSv9k 7.0.3.I7.3 and 9.2.2. In both OS I didn't found the command "system routing template-vxlan-scale". My vni peer doesn't come UP. Could you please suggest me on this?
I'm a bit rusty on Nexus now, but I think the template is ASIC specific, and may not work on a virtual image. If the peer is not coming up, check your underlay (routing, etc)
Hi hi thanks for your video. It’s really amazing but I have one small question for this video. Why do we need like any cast address because we are not going to use any other
When you configure the computer port, you do it as a regular access port. int eth 1/1 sw sw acc vla 1000 no shut My question is: Does this mean that when a computer/server is connected to that interface it will take the default STP timer until the forward state is taken? It is possible to configure the STP edge in that port or is not necessary?
Hi Sir, the example showing is 2 VTEP/Nexus9K directly connected. How if there's a middle node in between but not running VXLAN, is it required to enabled PIM on the middle node also? Thanks alot for the good video.
You videos are great. I have a small doubt here. You enabled system wide mtu as 9216 whereas interface level mtu is not modified. It will be 1500 by default. I did a similar setup. My nve peers were not showing. Once i modified mtu on interfaces to jumbo peers came up.
@@NetworkDirection Hello. Why you set mtu to over 9000? Vxlan overhead is 50 bytes for ipv4, you did it just for convenience? We encapsulate original frame (1526 - 4 bytes of original FCS) plus vxlan header (8 bytes), new UDP (8 bytes), new IP header (20 bytes), frame header and new FCS. So can I just set mtu to 1576? Base mtu of 1526 is correct?
@@linkflap1342 You definitely want to increase your MTU because of the various overheads, but it doesn't have to be over 9000. This just allows for the maximum frame size. From memory, I think this was recommended by cisco.
Thanks for making a great video :) I still have a doubt. What was the loopback address 3.3.3.3 used for on both the switches? Since 2.2.2.2 and 1.1.1.1 already is learned via OSPF and both switches have joined multicast group 230.1.1.1 they will receive the multicast packet.
Thank you for a lesson. I have done everything like was shown. The problem is that hosts are not pinging each other, moreover even those hosts that connected to the same switch, with the same vni configuration. I have checked many times the running config, searched the web, nothing useful was found. Can you advise please is there anything that can prevent L2 traffic from reaching between hosts ? Thank you in advance!
Sorry, I'm not sure what would cause this... The only thing I can think of is that the VLAN config on the ports is wrong? Or do the hosts have a host-based firewall that's blocking incoming ping? I think Windows will do that by default
Nice video. However, I am getting a bit confused here, as you had attached VLAN 1000 to VNI 5000. What if another customer needs to use VLAN 1000 considering multi-tenancy factor? Secondly, what if the same customer is using another VLAN, let's say VLAN 2000. Does it need another VNI or can it get attached to the same VNI?
Good questions. If you have two tenants, and they both want to use VLAN 1000, you could have them on different switches. On one, VLAN 1000 could map to VNI 5000. On the other it could map to another VNI, 6000 for example. Mapping a VLAN to a VNI is significant to the local switch. You could have 10 switches with a different VLAN on each, and they could all be in the same VNI. For your second question, if a single tenant had two VLANs, you would probably want to map them to different VNI's. Usually you would have two VLANs in the first place because you want to separate traffic.
@@NetworkDirection Hey , Please Explain above scenario in detail. 1. What exactly you mean by Tenant ? 2. How traffic flow will happen in case of tagging single Vlan to multiple VNIs? 3. Why it is required? Thanks in Advance!
1. A tenant is an entity that you want to separate. This could be customers, or business units. So, Customer A would be a tenant. Customer B would be a tenant, etc. 2. As far as I know, you can't tag a single VLAN to multiple VNI's on one switch 3. Not sure what you mean here. What is required?
there are two multicast group, one is RP multicast group, another is the multicast group for bum traffic. I mean what the difference between the two multicast group? what is the multicast traffic broadcast to RP? what is known and unknown multicast traffic?
Thanks Vijay. I have networking resources on Patreon (www.patreon.com/NetworkDirection ) and the website (networkdirection.net ) If you're specifically interested in VXLAN, there is a book you can check out: click.linksynergy.com/link?id=RL4E*8CmbSY&offerid=145238.2463561&type=2&murl=http%3A%2F%2Fwww.ciscopress.com%2Ftitle%2F9781587144677 (affiliate link, which means I get some of the proceeds if you buy it)
if we have 2 spines and 6 leafs for example, do we need ip pim rp-address xx.xx.xx.xx group-list xx.xx.xx.xx on leafs only and 2 statements ip pim anycast-rp x.x.x.x y.y.y.y on spines only, or anycast-rp on leafs as well ?
Would you share SPine config? I know your lab is with idea to simplifying Vxlan config, but in a Spine/Leaf design it will be a bit difficult to implement with SPINe out of the config? thank you.
I'm afraid I don't have a generic spine config to share. There's really very little VXLAN config on the spines anyway, They're just routers running an IGP, likely with a rendezvous point configured as well. They just forward traffic
I'm glad you like them! Part 6 is almost ready. It will be available in about 5 days time. When you subscribe, if you click the notification button (looks like a bell), you will be notified when it's released.
Very Nice explanation. I tried same setup and ping test between Laptop-1 and Laptop-2 is successful. Also ARP table is populated correctly on both hosts. But there is no output for "show nve peers" on either N7K. I even tried continuous ping test but still no output for "show nve peers". Is it any bug? I am using following N7K version: N7K-1# sh ver | i bin kickstart image file is: bootflash:///titanium-d1-kickstart.7.3.0.D1.1.bin system image file is: bootflash:///titanium-d1.7.3.0.D1.1.bin N7K-1# Note: I am using IOS router as Laptop-1 and Laptop-2. Laptop-1#ping 192.168.1.200 Type escape sequence to abort. Sending 5, 100-byte ICMP Echos to 192.168.1.200, timeout is 2 seconds: !!!!! Success rate is 100 percent (5/5), round-trip min/avg/max = 2/3/4 ms Laptop-1#sh ip arp Protocol Address Age (min) Hardware Addr Type Interface Internet 192.168.1.100 - 5e00.0000.0000 ARPA GigabitEthernet0/0 Internet 192.168.1.200 2 5e00.0001.0000 ARPA GigabitEthernet0/0 Laptop-2>sh int g0/0 | i bia Hardware is iGbE, address is 5e00.0001.0000 (bia 5e00.0001.0000) N7K-1# sh nve peers N7K-1# Thanks in advance
VXLAN bridging (aka flood-and-learn) does not have and layer-3 capability, so hosts will need to use an external router as their default gateway. BGP/EVPN (part 6 of this series) includes IRB (Integrated Routing and Bridging), where the Nexus switches can act as the default gateway
Have a look here: networkdirection.net/articles/routingandswitching/vxlanoverview/vxlanbridgingconfiguration/ That should include the info you need. Let me know if it doesn't
Great series of videos... However, you should not use IP addresses like 1.1.1.1, 2.2.2.2, and 3.3.3.3 in examples. Those are valid IPs. You should only use properly reserved addresses.
Hi Randy, thanks for your comment. Ofcourse you wouldn't use public IP's like 1.1.1.1 in the real world. 😀 I just expected for most people who are in VXLAN to understand that. I am more clear on it in my introduction to networking videos. Have a great day!
Just to share a gotcha, make sure you don't have a mac-address configured on the nexus interfaces that form the OSPF neighbors, spent 2 hours tearing my hair out as to why I was stuck in EXSTART on both switches.
