From: Dan Shechter (danshtr@yahoo.com)
Date: Tue Jul 20 2004 - 16:09:13 GMT-3
Thanks,
Is it over analyzing for the real lab? Or I should just configure it at
both end unless told otherwise, like "make sure the call won't collide"
Howard C. Berkowitz wrote:
> At 4:30 PM +0300 7/20/04, Dan Shechter wrote:
>
>> Hello all,
>>
>> I have read somewhere that the 'dialer load-threshold' should be
>> configured only on one end, because both side might initiate the
>> connection in the same time and the call won't succeed.
>>
>> Is it right?
>>
>
> Let me answer in a more general way. Anytime you have a
> connection-oriented function that can be initiated by either end, one
> of two conditions is necessary for reliable operation (i.e., without
> unresolved collisions)
>
> 1. The two ends have different trigger rules for starting the
> action,
> so it is highly probable that one will trigger before the
> other.[1]
>
> 2. The protocol (e.g., BGP session establishment) has a built-in
> mechanism for establishing requester-responder relationships in
> the event of the collision of simultaneous requests.
>
>
> [1] It will sometimes happen that, say, you set one side to trigger on
> 45% outbound link utilization and the other side to trigger on 65%
> outbound link utilization, and both ends just happen to hit the same
> trigger at the same time.
>
> The simplest answer to this is to observe "life is hard, and then
> you die." A more scientific answer says that well-designed functions
> have built-in collision resolution, as in #2 above.
>
> _______________________________________________________________________
> Please help support GroupStudy by purchasing your study materials from:
> http://shop.groupstudy.com
>
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sun Aug 01 2004 - 10:11:59 GMT-3