Question regarding mesh network and client-server communication

[WildCAt]

Hello!

I have a question regarding how mesh network "decides" how the request will be sent.

Imagine this situation: We have local network consisting of multiple routers(1-9), a server on one end and a client on another end.

Let's say, we have something like this:

I'm not a network specialist (just a software engineer), and I couldn't find any info describing the algorithm of how the hell request from client to sever (or the other way) is processed? Maybe I was using the wrong keywords...

I mean, server sends response > 1st router gets it. And then what? It goes like 1>2>5>7>9>client, decided by some algorithm? Or 1 sends it to all other connected nodes and client gets like ~18 responses and filters them?

Can someone explain it to me, please? Or provide any useful links?

 

Regards,

 

[Windows7ge]

I believe different hardware/software vendors use different methodologies so they're not all the same method but while I was taking a CISCO course on routing their implementation is based on "cost" a pre-determined value that a given interface or link is given. This number equates to being the "fastest route" to take to get a packet from A to B. The lower the cost, the faster the route and as so the route the router will send the data down because it's the fastest path.

 

There are many scenarios where there are exceptions to the rules such as ACLs, VIP traffic, any kind of black listing, VLANs, or if two routes have identical costs but at least for CISCO equipment this is how they do it. 

[Lurick]

6 minutes ago, Windows7ge said:

I believe different hardware/software vendors use different methodologies so they're not all the same method but while I was taking a CISCO course on routing their implementation is based on "cost" a pre-determined value that a given interface or link is given. This number equates to being the "fastest route" to take to get a packet from A to B. The lower the cost, the faster the route and as so the route the router will send the data down because it's the fastest path.

 

There are many scenarios where there are exceptions to the rules such as ACLs, VIP traffic, any kind of black listing, VLANs, or if two routes have identical costs but at least for CISCO equipment this is how they do it. 

Depends on the routing protocol for the most part. OSPF, EIGRP, RIP, BGP, etc. all use different cost mechanisms and have different ways to tweak preferred or number of equal cost paths as well.

Admin Distance and metric determine which route gets installed along with a few other things depending on if there is a tie on both of those and if ECMP is enabled/available as well.

[Windows7ge]

1 minute ago, Lurick said:

Depends on the routing protocol for the most part. OSPF, EIGRP, RIP, BGP, etc. all use different cost mechanisms and have different ways to tweak preferred or number of equal cost paths as well.

I was thinking of @ing you . This just further explains the complexity of how routing chooses which path to take. It's not one universal method for the whole Internet.

[WildCAt]

Thanks a ton to both of you! Now I have enough information to solve my problem and continue my research

[Windows7ge]

Actually now that I think about it. OSPF

 

Open 

Shortest

Path

First

 

It's been a while since I've been in the class but yeah EIGRP, RIP, BGP would all use a different method. Different vendors would also have their own methods but RIP and I believe OSPF are universal and allowed to be used by more than one vendor so other companies can use them.

 

If I recall EIGRP was developed by CISCO, it was proprietary but I've heard they're opening it up for other vendors to use.

 

BGP is very standard for connections between ISPs & homes/businesses but yeah all of them use a different method for getting packets from location A to location B.