Dumb Routing question

itdaddy

Okay. I do not know this answer on a granular level but I wanted to ask you guys the experts what is the truth about my question that I am going to ask.

Okay. I have 2 routers both have IP addresses in the range of 192.168.201.x say one is
192.168.201.1 255.255.255.255
and the other
192.168.201.2 255.255.255.255
but in router EIGRP 1 on both I advertise the network 192.168.201.0
These routers are located in different locations in this huge building of ours.
Is it efficient to advertise the entire subnet or should I only advertised the /32 host address only.
How does the router look at this type of advertisement does it know only to advertise the /32 host address
or does the traffic bounce around to location to location finding the correct address?
Maybe I should know this but I do not?

d4nz1g

Don't remember exactly regarding EIGRP, but in OSPF, when you use the network command, it does not advertise the subnet. The network command is used to match the interfaces which OSPF should be enabled.

Bardlebee

itdaddy wrote: »

Okay. I do not know this answer on a granular level but I wanted to ask you guys the experts what is the truth about my question that I am going to ask.

Okay. I have 2 routers both have IP addresses in the range of 192.168.201.x say one is
192.168.201.1 255.255.255.255
and the other
192.168.201.2 255.255.255.255
but in router EIGRP 1 on both I advertise the network 192.168.201.0
These routers are located in different locations in this huge building of ours.
Is it efficient to advertise the entire subnet or should I only advertised the /32 host address only.
How does the router look at this type of advertisement does it know only to advertise the /32 host address
or does the traffic bounce around to location to location finding the correct address?
Maybe I should know this but I do not?

Some CCIE's correct me if I am wrong, as I am a CCIE-in-waiting.

When you use the network command under the EIGRP process, you are not advertising the routes as you would be say in BGP. What you are doing is electing any interfaces with IP's in those segments to send out a MULTICAST hello message out every hello interval (5 seconds). This is on the multicast 224.0.0.10 address I believe for IPv4 in this case. If you were to use just a /32, you would get the same effect. So it would not be more efficient by any means to do so. Alternatively you can use the NEIGHBOR command to send out UNICAST neighbor hello's. In a point to point link I don't believe either of these would be more efficient then the others.

However, if you want to be safe about it, I suppose you could use the neighbor command or you could use a /32 command so that when you bring up other ports in the same network, it doesn't automatically try to find a neighbor. Or in another case, no one can plug into your router and try to create a neighbor as well. You can also choose to use passive-interfaces for this type of security as a better method though.

Hopefully that makes sense, and hopefully I am accurate.

d4nz1g

Bardlebee wrote: »

Some CCIE's correct me if I am wrong, as I am a CCIE-in-waiting.

When you use the network command under the EIGRP process, you are not advertising the routes as you would be say in BGP. What you are doing is electing any interfaces with IP's in those segments to send out a MULTICAST hello message out every hello interval (5 seconds). This is on the multicast 224.0.0.10 address I believe for IPv4 in this case. If you were to use just a /32, you would get the same effect. So it would not be more efficient by any means to do so. Alternatively you can use the NEIGHBOR command to send out UNICAST neighbor hello's. In a point to point link I don't believe either of these would be more efficient then the others.

However, if you want to be safe about it, I suppose you could use the neighbor command or you could use a /32 command so that when you bring up other ports in the same network, it doesn't automatically try to find a neighbor. Or in another case, no one can plug into your router and try to create a neighbor as well. You can also choose to use passive-interfaces for this type of security as a better method though.

Hopefully that makes sense, and hopefully I am accurate.

That was the long answer

10/10

Bardlebee

d4nz1g wrote: »

That was the long answer

10/10

Haha I do my written tomorrow, gotta throw out those useless facts see if I remember

itdaddy

thank you.

Iristheangel

Question/challenge for you: When issues like this arise, how do you solve it outside of asking on a forum? Did you try labbing it out in GNS3 first and seeing how it appears in the routing table? Not saying this to discourage you from asking questions but more of to challenge you to seek the answers through labbing out your questions and looking there first so you're forcing yourself to understand the output and how routing works.

itdaddy

irist, i only ask to talk to other guys, maybe i sound juvenile sometimes even though i have some certs, but i looked at route tables already and yeah i see what it does and I understand routing textbook wise just wanted some feed back sometimes....I don't really have anyone to bounce questions off sometimes and gets kind lonely by myself since I love this stuff...i have been known to ask some childish maybe questions, even my Physics teacher in college said i did but was getting a A- in his class when all else was either a C or failing...I have always asked questions that make me look weak in my education but kind of doubting Thomas sometime..but I totally get what you are saying...thanks for the feedback my friend.

..happy new year to you! by the way I think you have brocade certs what do you think of Brocade? are they similar to Cisco on the command line? do you like their stuff?

Iristheangel

It wasn't a criticism, itdaddy, or me saying you shouldn't ask questions. You're asking about traffic flow and whether you should advertise a loopback into EIGRP as a /32 or /24 and what would happen to traffic trying to go to one loopback or the other if you advertised a /24. This sounds like a fun one to spin up GNS and lab it out.

To answer your question, it's best practice to advertise it as a /32 (single host IP) instead of configured them in the same /24 subnet on different routers. You can do them in different /24 subnets on different routers but that's a waste of IP space for a loopback.

Bardlebee

I know those feelings itdaddy. At my job I am one of the senior guys, but I always ask a question immediately before googling. Just seems more natural to me to ask people instead of searching online or labbing it right quick. Of course, if I wanted more in depth knowledge I'll end up labbing it.

I know where you are coming from, definitely not juvenile.

itdaddy

Thanks Irist and Bard, yeah we use to advertise the /32 host ip only when i was at Fiserv data center using BGP of course but when I got here at my new job as network admin I saw a bunch of /24 on all these routers thought it might be inefficient practice...working in a data center with other network engineer breeds LOL makes you better engineer. My love is network engineering. I do a lot of stuff now but my goal will be in the NE field. I am a net admin now but will some day transition to maybe consulting since the $ are there and I can concentrate in my field of love choice hahahah but for now say 3 years here will be great since I am tasking with help setting up microave tower WAN connections and moving core switches and an entire data center to a new location. Once I have done that and had that in my belt, then Lord willing I can take my written CCIE but that is 3 years from now. I have a lot to learn but you will see me in the future with my CCIE. I wish I had as many as Iristheangel but that would be wishing my brain too much ) God bless you guys and happy new year..thank you for your feed back..always cheers mates!

itdaddy

iristheangel, wow i read your bio. how in the world did you get all those certs starting at 2008 only 7 years to burn all those? omgod you must be brainiac omgod the time to study...how can you dothat so fast. are you just a genius. i feel kind of stupid now haahhahaah
amazing feet and double ccie? wowowowowow

Iristheangel

I'm not a double CCIE yet Just working on it. As far as my certs, trust me - there are people that have done more in less time. Narbik Kocharians got his three CCIEs in a year and a half period if I recall correctly. Metha from LabMinutes got his three in the course of 18 months. Rick Mur had a few by the time he was 21 I think. I'm not special by any means.

As far as your goals, they're admirable and good and I really want you to achieve them. I can tell you right now that an expert isn't someone who knows everything. But to get to any of those guys (Narbik, Metha, Rick, Scott Morris, etc's) level, you got to dig into this stuff and immerse yourself. Read RFCs, lab out unfamiliar behavior, subscribe to blogs, read books, watch videos, etc. I bounce questions off people ALL the time when I hit a wall after trying everything and googling the hell out of it and sometimes you need a different perspective for sure. I'm not trying to get you to never ask questions - I'm trying to get you out of the mindset to ask first, research/lab/dig/Google second. It's natural when you're pre-CCNA or fresh CCNA to still not know the concepts but when you're goal is to move onto the CCIE and you're past CCNA and the CCNP ROUTE and SWITCH exams, you going to want to change your habits a bit. A lot of the time when you're working on your CCIE, you'll find things that don't behave how the official documentation says it would or you'll find documentation or books are straight up wrong about a concept and while you can ask people on a forum and either get their response and hope it's right or you can lab it out and find out yourself. When it comes to the CCIE lab, you're going to need to know those details inside and out on an intimate level. You can't get points you lose on the lab back even if you can prove documentation was wrong.

lostindaylight

Iristheangel wrote: »

A lot of the time when you're working on your CCIE, you'll find things that don't behave how the official documentation says it would or you'll find documentation or books are straight up wrong about a concept and while you can ask people on a forum and either get their response and hope it's right or you can lab it out and find out yourself. When it comes to the CCIE lab, you're going to need to know those details inside and out on an intimate level. You can't get points you lose on the lab back even if you can prove documentation was wrong.

So much this.

I liken it to hitchhiker's guide to the Galaxy. Tech notes and configuration guides can have errors, confuse you with irrelevant information while omitting important facts, ect. Sometimes you'll go to the command reference of all places and find something wonderfully well written and full of insight. Some RFCs are a great read, some are barely intelligible unless you are a programmer as it's mostly pseudo-code. You never know what you're gonna get.

A few months ago I watched a video series on BGP and the trainer was someone who's authored a couple of books that are considered must reads for a serious network engineer working with Cisco routers. His course was so riddled with errors I was stunned. In the long run it really pays to lab and verify everything, and not take what anyone says as fact.

Ok, rambling a bit now.

Anyway, good post, great point.

-lid