Solved: L4 protocols with Routing protocols

vpalania · ‎03-06-2006

Hi all,

I wish to know why OSPF is not using any L4 protocol(I know its encapsulated in IP),but see BGP is using TCP, also RIP is using UDP.But why OSPF is encapsulated in IP ?

Thanks,

Vijay

pkhatri · ‎03-06-2006

Vijay,

1. In Unix-like systems, you need to have a process owned by the root user in order to send out packets which use a L4 protocol other than TCP/UDP - not sure if that is true anymore but that was the case when OSPF was designed.

2. ISIS was designed for routing CLNP (Connectionless mode network protocol) packets, which is a layer 3 protocol with the ISO suite. This provides even more security than IP. Since it's run directly over L2, you cannot route the ISIS protocol packets. Therefore, it's hard to inject malicious packets into a segment from a remote network. The CLNP protocol uses NSAP addresses.

Hope that helps - pls rate the post if it does.

Paresh

View solution in original post

pkhatri · ‎03-06-2006

Hi Vijay,

A good source of this information is John Moy's 'OSPF: Anatomy of an Internet Routing Protocol'

Here's a quick summary:

- OSPF did not require the reliability of TCP. Since link-state routing protocols have reliablity built into their flooding algorithms, the OSPF designers felt that TCP would "just get in the way"

- because it is easy for most users to send UDP packets on Unix-like systems, but the sending of packets directly over IP requires special privileges, it was felt that running it directly over IP would provide OSPF a bit more security.

- the UDP header would also have added another 8 bytes of overhead to OSPF packets

Hope that helps - pls rate the post if it does.

Paresh

vpalania · ‎03-06-2006

Hi,

Thanks for your information.

I have two more questions:

1.Can you explain more on the security difference in transmiting IP and UDP/TCP ?

2.Also, ISIS is a Link State Routing protocol, then why it is encapsulated in L2 instead of any L3, I am not sure which L3 supports NSAP address ?

Thanks,

Vijay

pkhatri · ‎03-06-2006

Vijay,

1. In Unix-like systems, you need to have a process owned by the root user in order to send out packets which use a L4 protocol other than TCP/UDP - not sure if that is true anymore but that was the case when OSPF was designed.

2. ISIS was designed for routing CLNP (Connectionless mode network protocol) packets, which is a layer 3 protocol with the ISO suite. This provides even more security than IP. Since it's run directly over L2, you cannot route the ISIS protocol packets. Therefore, it's hard to inject malicious packets into a segment from a remote network. The CLNP protocol uses NSAP addresses.

Hope that helps - pls rate the post if it does.

Paresh

arvindchari · ‎03-06-2006

OSPF uses multicast addresses 224.0.0.5 (All OSPF Routers) and 224.0.0.6 (All OSPF DR's & BDR's).

Also to my best knowledge, RIP v1 uses a broadcast address of 255.255.255.255 and v2 uses a multicast address of 224.0.0.9.

As for why it uses IP, I think the term "the best network layer protocol in the world" says it all :D

HTH

Arvind