Logical control of routing?

Artos

I'm currently doing a course in Networking and I've just come across a question which may as well be in a foreign language to me. If anyone has any ideas they would be most appreciated. If this has an obvious answer, don't mock me - i haven't really gone to any lectures...:cool:

Question:
Consider the needs for the logical control of routing in networks(specifically in packet switched networks) - what simple rules would you propose for effective network operations?

Cheers.

rmacm

Artos wrote: »

i haven't really gone to any lectures...:cool:

Going to them might be a good idea :rolleyes:

Artos wrote: »

Consider the needs for the logical control of routing in networks(specifically in packet switched networks) - what simple rules would you propose for effective network operations?

As to your question try give a google around and look in the FAQ sticky at the top of the forum. The Cisco website has some good information although it is very Cisco centric. If I'm looking at the question in the right way one of the immediate things that springs to mind would be to keep routing tables as simple as possible. This would improve routing performance in a network.

FusionNet

Question:
Consider the needs for the logical control of routing in networks

No offence but maybe you should consider a different course, this one obviously doesnt float your boat....!!!!

Artos

I would, but the exam is in 2 weeks and its too late now to drop. Besides it fit in my schedule. Gotta keep tuesdays free for golf and whatnot :rolleyes:

degsie

Artos wrote: »

I would, but the exam is in 2 weeks and its too late now to drop. Besides it fit in my schedule. Gotta keep tuesdays free for golf and whatnot :rolleyes:

Sounds like you would be a perfect candidate for a customer service role in Eircom

monkey tennis

Pretty vague, open-ended question TBH, unless there's a context that was built up with other questions.

rmacm wrote: »

keep routing tables as simple as possible. This would improve routing performance in a network.

Nope.

rmacm

monkey tennis wrote: »

Nope.

Was that a nope as in it being incorrect if so why?

Cheers
Rory

TimTim

rmacm wrote: »

Was that a nope as in it being incorrect if so why?

Cheers
Rory

Having just recently done a ccna course (waiting to do my exam) nope as in being incorrect.

I'll try explain with a really poor diagram

Router 1 -------------> 56k line -------------> Router 4
|                                                                      |
|                                                                      |
| 100Mb line                                                       | 100Mb line
|                                                                      |
|                                                                      |
Router 2 --------------> 100Mb -------------> Router 3

If you are at router 1 and you want to get to router 4 and you want to keep things simple with your routing protocol, then you'll travel over the crappy 56k line and get ****ty performance.

If you have a somewhat intelligent routing protocol, it will take more than simplicity into consideration and use other metrics to determine the best route, so you'll travel over router 2 and 3, but your performance will be a million times better (If anyone gets pedantic over that, take it up with someone who cares)

rmacm

TimTim wrote: »

Having just recently done a ccna course (waiting to do my exam) nope as in being incorrect.

I'll try explain with a really poor diagram

Router 1 -------------> 56k line -------------> Router 4
|                                                                      |
|                                                                      |
| 100Mb line                                                       | 100Mb line
|                                                                      |
|                                                                      |
Router 2 --------------> 100Mb -------------> Router 3

If you are at router 1 and you want to get to router 4 and you want to keep things simple with your routing protocol, then you'll travel over the crappy 56k line and get ****ty performance.

If you have a somewhat intelligent routing protocol, it will take more than simplicity into consideration and use other metrics to determine the best route, so you'll travel over router 2 and 3, but your performance will be a million times better (If anyone gets pedantic over that, take it up with someone who cares)

Will be doing the CCNA shortly (or rather finishing it) myself. I've been training with work and we've been covering route aggregation and how it can make routing tables more efficient although the point is pretty much negated with the more intelligent routing protocols. I suppose the point I was trying to make although I didn't explain it was that you'll use less bandwidth during advertisements, less cpu on the router when calculating routing tables although doing this could lead to sub optimal routing like in the case above.

Aside from that modern routers generally don't use the whole routing table for making forwarding decisions they use a forwarding table which should contain the routing protocols preferred routes for forwarding packets so as you mentioned the forwarding table should contain the route with the quicker links. Anyway I must go and have a few pints it's been a long week.

Cheers
Rory

TimTim

Yeah I can see where you are coming from in a way.

Although if you are running into CPU/memory issues on your router during updates then you've bought the wrong router (although I'll admit some level of aggregation is a good thing).

For internal networks of less then 100 *cisco* routers then use EIGRP. In 100+ /mixed vendor environments then use OSPF.

If you are going outside then you should use BGP.

Capt'n Midnight

You could cheat and let the network do all the routing dynamically for you
http://en.wikipedia.org/wiki/Open_Shortest_Path_First
http://en.wikipedia.org/wiki/Border_Gateway_Protocol

keep routing tables simple
if possible have physically close sites on close subnets so you can subnet them together when doing rules elsewhere

set the main gateway as the the one with most rules so you only have to list the exceptions even if most traffic goes that way

default gateway to internet = 10.1.0.1
gateway to WAN = 10.1.0.2

if you set the internet gateway as the default then you have one rule for all the WAN offices no matter how many there are
one route on the internet router to point to 10
and a second route on all clients so they the wan works even if the internet router goes down
route -p add 10.0.0.0 mask 255.0.0.0 10.0.0.2

rmacm

TimTim wrote: »

Although if you are running into CPU/memory issues on your router during updates then you've bought the wrong router (although I'll admit some level of aggregation is a good thing).

Agreed the only time I've seen routers being over burdened by large routing tables is when older equipment is in use.

TimTim wrote: »

For internal networks of less then 100 *cisco* routers then use EIGRP. In 100+ /mixed vendor environments then use OSPF.

If you are going outside then you should use BGP.

I'd go for OSPF over EIGRP given that it's a Cisco proprietary protocol and yes between autonomous systems BGP should be used.

merkuree

Originally Posted by TimTim
Although if you are running into CPU/memory issues on your router during updates then you've bought the wrong router (although I'll admit some level of aggregation is a good thing).


Also a function of how many routing protocols you are running on the router.....and any redistribution occurring......plus other functions occuring on the router - extensive ACLs, firewall, NAT.....should not occur but you see it in the field.

In some cases, it may be related to the function that your router plays....designated router in OSPF etc etc.

Sometimes, just bad design and poor execution.