11-22-2009 12:36 PM - edited 03-04-2019 06:46 AM
Hi
I'M currently studding my BSCI Certification Exam, currently I'm studdiyng on subject about BGP,
I understand the necessity to using a loop back interface to established neighborship when multiple path exist between 2 BGP speakers, but I do not understand why we would always make the local router as the next up for advertized routes in thoses cases.... why the default next up should not be used ?
Also, in the "
Understanding and Troubleshooting BGP Neighbor States" section, the author explain the different neighbor state, I confuse about the Establish state when she wrote :
"If a response does come back in a timely manner, BGP goes to the open confirm state and starts scanning (evaluating) the routing table for the paths
to send to the neighbor. When these paths have been found, BGP then goes to the established state and begins routing between the neighbors."
Why the router should check if it have the route to this router if the TCP connection is already open in "Open confirm" state ? IT's maybe my English, but something is not clear to me in this... !
Thanks a lot for your halp in advanced !
Solved! Go to Solution.
11-22-2009 03:07 PM
Hello,
Regarding your first question, I assume you are asking whether we are supposed to use next-hop-self when also using the update-source to peer two BGP neighbors. The fact is that these commands are completely independent and you may at any time use one or the other, or both of them. However, in many situations, the next-hop-self together with update-source using loopbacks simplifies a couple of things because the loopbacks are usually already advertised in an internal routing protocol and thus they are known. In such case, it is not necessary to advertise the links to external BGP neighbors into the internal routing protocol. Using both these commands in configuration of a BGP neighbor leads to more conveniently-looking routing tables and to a slight simplification of internal routing protocol configuration. But once again, it is not mandatory to use both of them, it's just more convenient.
Regarding your second question, is the quotation from the book exact? If yes then the book is a mess. There is no relation between "scanning for paths to send to a neighbor" and transitioning to a different BGP state. Without talking too much about BGP states right here, I suggest you read the following URLs and then come back with more detailed questions:
http://archive.networknewz.com/networknewz-10-20060403BGPAdjacencyStates.html
http://ccnprecertification.com/2005/05/31/bgp-state-machine-timers-and-messages/
http://en.wikipedia.org/wiki/Border_gateway_protocol#Finite_state_machine
http://www.informit.com/library/content.aspx?b=CCIE_Practical_Studies_II&seqNum=79
Best regards,
Peter
11-22-2009 03:07 PM
Hello,
Regarding your first question, I assume you are asking whether we are supposed to use next-hop-self when also using the update-source to peer two BGP neighbors. The fact is that these commands are completely independent and you may at any time use one or the other, or both of them. However, in many situations, the next-hop-self together with update-source using loopbacks simplifies a couple of things because the loopbacks are usually already advertised in an internal routing protocol and thus they are known. In such case, it is not necessary to advertise the links to external BGP neighbors into the internal routing protocol. Using both these commands in configuration of a BGP neighbor leads to more conveniently-looking routing tables and to a slight simplification of internal routing protocol configuration. But once again, it is not mandatory to use both of them, it's just more convenient.
Regarding your second question, is the quotation from the book exact? If yes then the book is a mess. There is no relation between "scanning for paths to send to a neighbor" and transitioning to a different BGP state. Without talking too much about BGP states right here, I suggest you read the following URLs and then come back with more detailed questions:
http://archive.networknewz.com/networknewz-10-20060403BGPAdjacencyStates.html
http://ccnprecertification.com/2005/05/31/bgp-state-machine-timers-and-messages/
http://en.wikipedia.org/wiki/Border_gateway_protocol#Finite_state_machine
http://www.informit.com/library/content.aspx?b=CCIE_Practical_Studies_II&seqNum=79
Best regards,
Peter
11-22-2009 05:21 PM
Hi Peter,
I read the first 2 articles the first one is which one is most corresponding to my understanding based on what I had read in my Cisco Press book, but the second one is in conflict with the first one about the active state.
My understanding in the first article the active state is meaning a problem in the configuration on one of both router, in the second article the active state is a state in which all router will be in stage of becoming "establish" (It's maybe my English, I know my English is defenitly not perfect, but it's what I understand)
at this time I did not take the time to read the two other article you had provide, but I will the time for sure to read them
For fhe first question, your explanation make sence to me !
Thanks a lot !
11-22-2009 05:35 PM
Hi !
Sorry for very bad rating, it was'nt I would like to set, is it possible to changed after it saved ?
also I missed maked the discussion as closed is'it possible to reopen it ?
thanks a lot !
11-22-2009 08:02 PM
It is because the new rating is different then the old one and people can easily give some one a 1 when they actually meant 5. It takes some times to get familiar with it.
I accidentally did the same thing for some one else.
I will give Peter a 5 to make up for yours.
Also, you can continue with this one or open a new one
Reza
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide