Why BDR is elected first?

[Balajee Muggalla]

Hi,

   I tried searching the internet but did not find any relevant discussion. Anyone know why BDR is elected first and then it is promoted as DR? DR can be elected first and then BDR na?

Thanks

[Flavio Miranda]

Hello @Balajee Muggalla

If you read the RFC 2338 Here  there´s an statement that says:

"The reason behind the election algorithm's complexity is the
desire for an orderly transition from Backup Designated Router
to Designated Router, when the current Designated Router fails.
This orderly transition is ensured through the introduction of
hysteresis: no new Backup Designated Router can be chosen until
the old Backup accepts its new Designated Router
responsibilities."

By making they pass through  the process of BDR first, ensure an order on the DR election process.

-If I helped you somehow, please, rate it as useful.-

[Balajee Muggalla]

Hi,

   Thank you for your reply. Still my question is not clear. How moving BDR->DR ensures orderly transition? Why not vice-versa?

Thanks

[paul driver]

Hello

The DR is responsible for sending/receiving routing updates so if the DR fails the BDR needs to be promoted to DR status so it can continue servicing these updates.

These updates are seen via two specific multicast addresses:

DR - 224.0.0.5
Spokes - 224.0.0.6


res
Paul

[jobraswe@cisco.com]

Very logical, well documented and referenced response... thanks!!!

[Tinashe Ndhlovu]

Hi,

i will I'll try to answer you based on my understanding of the election process and not based on any official document as to why this happens.

There will only ever be one DR and one BDR in a broadcast segment. And you want only the DR to speak the BDR and the rest to listen. So by electing the BDR first, you avoid a situation where the DR starts sending info to a router that still thinks that it is the DR. Remember unless you change the DR/BDR flag, routers at election time will send a message that says "I am the DR". Now imagine during election a router starts receiving info from the elected DR but as far as it is concerned it is still the DR.. so we elect the BDR first and tell that Router sit down and be humble.. then we elect the DR and all is well.. 

again i I am writing this off the top of my head so I could be wrong but I believe that is why during the election process we elect the BDR first because that tells the elected BDR Router keep quiet.. because all routers unless you fiddle all think they are the DR and they are dying to talk and flood route info.. 

[Richard Burts · Accepted solution]

The original poster asks a question that we have not yet answered " How moving BDR->DR ensures orderly transition? Why not vice-versa? ". It seems to me that the answer to this is that the existing implementation has a single election process. When we run the election code we are always electing the BDR, and once elected as BDR the router is eligible to become DR (but that does not require an election process). If we did it vice-versa then we would have two election processes. There would be an election process to elect DR and a different election process to elect BDR. 

HTH

Rick

[paul driver]

Hello

Thanks for the clarity Rick - I thought I had answered it when i mentioned its needs to be a DR before sending/receivng updates now reading you post I guess I didnt.

res
Paul

[Balajee Muggalla]

Thanks for the clarification, Rick.

[Richard Burts]

You are welcome. I am glad that my explanation was helpful. Thank you for marking this question as solved.

Paul

You did talk about the importance of the DR and the fact that if the DR fails then the BDR is promoted to DR and this is an important part of the answer. But it did not get into the election process and did not address why BDR is elected first. I thought it was important to make the point that with the current implementation where BDR is elected first that one election process is enough. But it you elected DR and then elected BDR that it would require two different election processes and this would be more complex.

There is another aspect of the election process which I started to mention in my previous explanation but decided that it made the post more complex. Now that we are looking at that question again I believe that it might be helpful to point out this other aspect, which helps explain why BDR is elected first. Let us start with a quick review of the purpose of the BDR which is to be able to take over if the DR fails. To accomplish this the BDR needs to know about the network just as the DR does and so the BDR receives and processes the messages to 224.0.0.6 (which we typically think of as the address for the DR). So the BDR has the complete information that the DR does and if the DR fails then the BDR is ready to take over (and there is no delay while the new DR gets up to speed). But think about what would happen if the DR were elected first and there are separate election processes for DR and for BDR. With separate election processes if the DR fails then OSPF would need to run the DR election process. But that raises the possibility that some router other than the BDR might win the election and become the new DR. But in this case the new DR would need to learn all the information which has been sent to 224.0.0.6 before it could effectively function as DR.

By having a single election process for BDR it is easy to assure that when the DR fails that the BDR will immediately be promoted to DR and begin to function without delay. Then a new election will be held for BDR which is much less disruptive than having an election for DR.

HTH

Rick

[jobraswe@cisco.com]

Best overall in depth explanation I have seen... Very Good!!! tested and watched the debug messages... the BDR was in fact elected first! Thanks!!!

[Richard Burts]

Thank you for the kind words. This was an interesting discussion about a subtle aspect of the routing protocol. And it is a good example of what the community can accomplish when several people provide their insights into answering a question.

HTH

Rick