|
Home > Archive > CCNP > July 2002 > Router ID selection in OSPF
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Router ID selection in OSPF
|
|
| sukuvi 2002-07-13, 11:47 am |
| Hi Everyone,
I have searched on the Cisco site but couldn't find an answer.
We know that for the router ID, the highest IP address is selected whether or not the interface is participating in OSPF. A loopback interface IP address is selected instead if one exists.
What if there are many loopback interfaces, such as
loopback 0, ip address 1.1.1.1
loopback 1, ip address 172.1.1.1
loopback 2, ip address 220.1.1.1
loopback 3, ip address 20.1.1.1
Which one will be the router ID?
Is the router ID selected based on the IP address or the loopback interface number? | |
|
| Hi
The Cisco Press Routing Guide does not give a definitive answer either. However, as you correctly state the router ID is determined by the highest IP address, unless a loopcback interface is configured. In that case, the loopback IP address is used as the router ID.
Why anyone would want more than loopback interface is beyond me, after all the loopback interface is virtual, not physical and can never go down, so it is always available.
Extending the theory of 'highest IP address' to the loopback interface, I would guess that the highest loopback interface IP address would be chosen.
Anyone else?
Hippo | |
| sukuvi 2002-07-14, 10:48 am |
| I was playing with the Practice Test on the CD that accompanied the book "CCNP Routing Exam Certification Guide" by Claire Gough ISBN 1587200015, (the routing book from the CiscoPress CCNP Certification Library) and came across a similar question:
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Which of the IP addresses on the following interfaces used on the same OSPF router would be chosen as the OSPF router ID?
A. Loopback1: 172.19.100.10
B. Serial0: 192.168.100.1
C. Loopback0: 172.30.10.1
D. Ethernet0: 10.100.10.1
The answer is C.
~~~~~~~~~~~~~~~~~~~~~~~~~~
One negative aspect about this practice test is that, there is no explanation about the correct answer on the test itself, the way you would see it in other practice tests such as Trancender or Boson, but rather a reference to a chapter, subtitle in a chapter, or even a section in the book.
For this particular question, it refers to "OSPF Network Topologies", which I found in Chapter 5: Using OSPF in a Single Area, on page 212.
But under this subtitle there is nothing about the Router ID selection. The next subtitle, "The Hello Packet", has a table called The Hello Packet (Table 5-3) on page 215, where there is reference to the Router ID Selection concept but there's no mention of cases where there might be more than 1 loopback interface.
As we can see, Loopback1 IP address is higher than Loopback0's. So my conclusion is that, when it comes to selecting an ID from many loopback interfaces, the one with the lowest number wins. This is why Loopback0 is the correct answer. My conclusion will be valid only and only if there's no bug in this practice test. | |
| rsteedma 2002-07-14, 7:16 pm |
| Hippo's Theory is right! 
Router ID is selected by this process:
(1) A manually configured router-id
(2) The highest loopback IP address
(3) The highest physical IP address
C is correct on the practice test because 172.30.10.1 is the highest loopback IP address configurged, NOT because of the "0" in the loopback interface name.
Here's a URL with more details:
http://www.cisco.com/warp/public/103/eigrp-dup-id.html
>>>>> http://www.congonetworking.com/ccnp.htm <<<<<
Free CCNP Message Board & Practice Tests - Drop on by!! | |
|
| My theory was based on an educated guess. I am using Claire Gough's CCNP Routing Guide (Cisco Press), and although it does describe the router ID selection process, is does not cater for the situation where there is more than one loopback address.
So, I still have an unanswered question; why would there be a need for more than one loopback address?
Hippo
 | |
| rsteedma 2002-07-15, 3:31 pm |
| One loopback should be enough for stability and load balancing purposes.
Multiple loopbacks are useful for simulating fake networks in a lab or test environment.
They can also be useful when you decide to migrate from two routers into one. You can "simulate" the existence of both. Secondary addresses could also be used.
Hmm... any more ideas?
Cheers,
Ron
>>>>> http://www.congonetworking.com/ccnp.htm <<<<<
Free CCNP Message Board & Practice Tests - Drop by!!! | |
| sukuvi 2002-07-16, 9:18 pm |
| Rsteedma,
Are you saying that IP address 172.30.10.1 is higher than 172.19.100.1?
If that's what you are saying, how do we calculate which one is higher? | |
|
| 172.30.10.1 is NUMERICALLY higher than 172.19.100.1
Start with the leftmost octet; 172 is the same for both addresses, then the next octet 30 is NUMERICALLY higher than 19. So 172.30.10.1 is the highest address.
OK
Hippo | |
| SureshHomepage 2002-07-18, 5:06 am |
| Hippo was right. It is 172.30.10.1, the highest IP address than 172.19.100.1.
The order is like Router ID--> Highest Loop back address--> Hishest Physical Intf Address. | |
| Always Learning 2002-07-18, 9:25 pm |
| Hippo...there are many needs for more than one loopback address. It really depends on what you are trying to do. One purpose that comes to mind is setting an interface to do NAT overload. | |
| rsteedma 2002-07-19, 1:14 pm |
| Could you be more specific about a NAT scenario you encountered where MULTIPLE loopbacks are either required or simplify the configuration?
Cisco NAT Configuration Webpage:
http://www.cisco.com/warp/public/556/12.html
I've only configured NAT a couple times, but I didn't need to use multiple loopbacks. I'm curious.
Always learning is right...
Cheers,
Ron
---
Ron Steedman CCNP, CCNA, MCP, BS Computer Science
Free CCNA Practice Tests, Study Guides, & Message Boards!
http://www.congonetworking.com/ | |
| Always Learning 2002-07-19, 4:19 pm |
| Lets say you want to nat your internal network to a public ip address for internet access. One solution would be to do nat overload and you could configure a loopback int for the nat overload interface. On the same router lets say you want to monitor it. Configuring a loopback interface to monitor is much more reliable than monitoring a phyiscal interface, especially if you have redundant links to the same place.
Like most things there are more than one way to skin a cat. This is just one example that I could think of.
Does that make sense or am I off my rocker??? | |
| cahillrobert 2002-07-22, 7:15 pm |
| I came into an organization that used a private sequential number for network monitoring.
Later when we upgraded the network I choose to use another Lo numbering scheme to optimize OSPF. So they ended up with two Lo addresses.
The numerically higher Lo does win. Proved in a lab to some of my co-workers several times. Reminder after adding/changing a Lo address you will probably need to reload the DR and BR. If you want to immediately view the changes made.
- Bob | |
| mikeghet 2002-07-22, 11:10 pm |
| If the dr or bdr goes down and has a physical address a new dr or bdr elecetion will begin. You may have this router strategically placed and may not want a new dr/bdr elected. So even if the router is not working the logical ip address never goes down and that way a new dr/bdr is not elected. This may come in handy on a network that does not converge. |
|
|
|
|