FullMikroTik MTCRE - Failover, Ptp Addressing, Interface Based Routing (Episode 6)

[@TheNetworkBerg]

NB - Interface based routing only works for Ptp configurations, trying to do this on a link where multiple hosts can reside will not work. Reference Material: wiki.mikrotik.com/wiki/Manual:IP/Route help.mikrotik.com/docs/pages/viewpage.action?pageId=59965493

[@mishasawangwan6652]

hey @The Network Berg : i have 2 rb5009s and a single ISP who has allocated me a block of 5 static ips. i’ve configured 1 static ip on each of the rb5009s which each have their default GW pointing to the ISP gateway. the ISP provided modem is in “bridge/passthrough” mode so it doesn’t do anything except provided access to the internet. my setup so far works fine (1 subnet for management and another subnet for all lan) but am wondering if there’s something i could be doing better. not much info on this on the net. any thoughts? cheers.

[@gyorgykovacs6781]

You infected me, I fell in love with Mikrotik and routing. ;) Your passion to routing is impressing me. I had the same feelings when i started to learn programming(C, C++, Java) 6 years ago. It’s an inexplicable feeling when we see behind things and put together the details and understand how things work. You helped me to begin to understand this networking world as well. And thank you for your selflessness in sharing your valuable knowledge with us. People like you make this world a better place

[@diwasbhatta5951]

such a underrated youtube channel..this channel desearves to be very famous within IT students

[@kresimirpecar4925]

If we have for eg. vDSL router for ISP1 and LTE router for ISP2, if we want to have ISP2 as an failover (no PPPoE), we should make ether2 WAN, so ether1 is WAN1 and ether2 is WAN2, we asign them IP addresses, then remaining ports go to bridge and we enable dhcp server for that bridge. Then we can make static routes to WAN1 and WAN2 IP addresses, put distance for WAN1 to 1 and 10 for WAN2. Also we need to masquerade for each WAN IP so if traffic want to leave via WAN1 it will be NATed for WAN1 IP, and if it wants to leave via WAN2 it will be NATed for WAN2 IP. I hope i understand correctly. Great video as always, short so you can keep someone attention but educational. Maybe it's better to have more shorter videos than few longer ones.

[@TheNetworkBerg]

You are totally correct :D Creating videos is a fine line, sometimes a video is too short then doesn't get promoted by KZbin, but on the other side a video may be too long and then becomes too frustrating to watch in one sitting. I am trying to keep a balance between things. A typical rule of mine is to aim for 15 mins in a video, although sometimes I just get carried away and exceed this limit. It happens more so in the MTCRE topics as some of these things you need to talk about in one video.

[@geoffvandermerwe6220]

Thanks keep the videos coming !

[@TheNetworkBerg]

Still have plenty to come 😀

[@Rob_Turner_UK]

Great explaintion

[@soroushsarkari9977]

Thanks for your wonderful channel and videos 🙏🏻

[@TheNetworkBerg]

Thank you for the kind words!

[@Vinothmanick]

Great video as usual kudos ✌🏻😎

[@TheNetworkBerg]

Thanks 👍

[@salmanseed1207]

God bless you

[@IronTrela]

I like your videos so much. Which software you use for network design at the beginning of the video? Keep up the good work!

[@TheNetworkBerg]

It's a web tool (draw.io) can create free diagrams on it :)

[@alimibrahem8120]

Very very thanksful man, you are a wesome..!. However, the amazing thing that i wounder how you do is how you can combin the eve topology with the winbox..? So when you disconnect the line the winbox session is quickly response to that..? Please can you tell us how we can do that..?

[@jasonvanwyngaard4387]

Thoroughly enjoy your channel. Seen very elaborate failover setups using mangle. Why use that instead of what you've demonstrated?

[@TheNetworkBerg]

There are hundreds of ways to implement failover, it depends on the requirements and how you want things to failover in the event of an error. Personally as an ISP I prefer having BGP configured to CPEs and then having that protocol act as a failover mechanism. I've even seen some people do things with Netwatch to disable a primary route in the event a failure. Sometimes though a very simple solution like in this video works fine for failover

[@biki1973]

what if i have server behind my router, when main ISP is down server will be visible under different ip from second ISP, is there any clever way to make it work?

[@TheNetworkBerg]

I assume you are talking about external access to the server. So if the primary ISP goes down, how can people still connect to the server when its public IP changes. Well... I think the easiest solution for that would be to implement DDNS (Dynamic DNS). So that if the primary does go down the DDNS can be updated automatically to the new public IP which you will also need to add NAT rules for. Otherwise you can manually update an A record to the new public IP, but that can have its own issues especially if the address you receive may be dynamic. So I really think DDNS is the best option for a scenario like that.

[@DimitriPappas]

Great content as always. I do think it would be awesome if you could cover Ptp addressing in a bit more thorough detail at some point, as you barely skimmed the surface of that topic enough for anyone to really grasp how to implement it properly, and to be honest I don't quite understand the implementation you showed (with 100.64.0.253 as an address and 100.64.0.1 as the network/peer - these IP's are not adjacent with each other, so how does this even work exactly?). Could ANY IP be used as the network IP even if it shares nothing in common with the primary IP address? I also didn't know about /31's until now, or that we can even use /32 with peer as network address to establish a link. Until now I've just used /30's but that seems a bit wasteful for cases where a /31 or /32 configuration would work just fine. Your video did prompt me to try read up a bit more about this on the mikrotik forums for some answers, but I feel like this is still quite an uncomfortable/grey area. It it seems /31's are actually discouraged because of incompatibilities between different vendors/standards, particularly when it comes to dynamic routing protocols like OSPF, and surprisingly it seems Mikrotik are somewhat behind with their implementations/compatibility with /31 and /32 ptp which surprises me a bit given their prominence & speciality in the field of routing. Hoping to therefore hear your thoughts & sentiments on this in more depth and whether you think it's even worth using /31 and /32, especially in production/enterprise environments? Cheers! :)

[@TheNetworkBerg]

Not sure why KZbin labeled your content as spam, might be because of the IP addresses since they are highlighted for some reason. But this made your comment hidden until I had to go into a spam section and approve it. Correct, some vendors do not support /31 addressing at all which is why it is more normal to see /30 addressing being used for most point-to-point links. If you are going to use /31 with another network first confirm with them if their equipment supports it. I think it's fairly straight forward to see PTP addressing in action whenever a client forms a connection to any of the PPP services, you will see that the MikroTik server will give itself an IP, and that a remote IP will be given to a client where these two devices can directly communicate. In MikroTik's case with using the Network address as the adjacent device's IP it is not required for this IP to be in the same pool. You could have your IP as 1.1.1.1/32 and the Network address as 2.2.2.2, if the remote side has its IP as 2.2.2.2/32 and the network as 1.1.1.1 then both devices will be able to communicate on L3 without any issues. Although it does make logical sense to add IPs in a sequence as that makes sense to us as we've been doing it for years this way. So you could use 1.1.1.1/32 and 1.1.1.2 as a network (heck can even start at 0 or end at 255 that's what makes ptp nice) On our network we actually make use of PTP addressing quite extensively on our L2TP tunnel connections, although it's more in a static sense as we do not use pools for our clients as we want each client to have static IP bound to it so we tend to just assign 2 IP addresses out of a reserved /22 pool. One IP for the server and then one for the client. We also make use of PTP addressing on a few of our BGP peers, and it has been working correctly without any issues for the last 3 or so years. So yeah, if you want to save on some IPs by using /31s for stuff like PPP services then I could recommend it from personal use.

[@DimitriPappas]

@@TheNetworkBerg Thanks for coming back to me on those points with such a great reply. Super interesting stuff :) I have indeed seen that ptp phenomenon with pppoe connections before (even years ago with ADSL) and always wondered how that was possible. Now we know :D I guess I'll be sticking to /30's for anything serious for the time being as that's guaranteed to work across the board, but it's good to know about this stuff and definitely something to play with in the lab to see where it could potentially be used 👏🏻

[@matej_stepan]

The hardest thing about failover is to determine it. Check gateway is not a great option for internet use because it checks the gateway not the internet. I have several IPs just for check fail purposes.

[@TheNetworkBerg]

We'll take a look at recursive routing in the next video which is very nice for internet based failover scenarios if the gateway remains up even if the "internet is down"

[@ashleygraemetari523]

@@TheNetworkBerg perfect. Your tutorials are awesome as always.

[@DimitriPappas]

I use custom scripts & scheduler to handle most of my Multi-WAN failover, because as you say the ping check doesn't always work. Just because the gateway is reachable/pingable, doesn't mean the breakout to the internet side of things is operational. So it's best to ping a real world public IP and make a determination based on that

[@berry9312]

can we send bandwidth from city A to city B with mikrotik..

[@TheNetworkBerg]

Definitely, you can even implement stuff like OSPF, BGP or Policy Based Routing and send traffic to as many cities as you want to.