Just for my person education and understanding, BGPd for example use the
lists functions from the system instead of any of the various DB functions.
Obviously the lists functions are kept in RAM, so this is definitely
much faster, a lots, but when is it that one would switch to DB
functions instead of lists. What kind of rules of thumb would one use to
select one over the other?
I guess is the routing table were 10 millions routers, it may simply
doesn't make any sense as the memory might well not be there to support it?
What is the situation where it makes sense to use the built in DB
functions and witch one to elect using instead of the lists?
I am trying to understand when it makes sense to select what in the
design time.
My situation now is about 350K records and might grow to may be 500K.
But technically, in the maximum limits, I can't see a case where the
operation would be so big as to use that much, but it's always possible
to reach 500 millions records, but I can't imagine that possible really.
I was and am thinking to use SQL instead as it make sense, but in
practice, 300K records should really covers it all.
So, what's logical to use here you think?
If BGPd had to carry let say, 10 millions routers, would the design have
been the same for the selection of lists as it is now?
What would have make it different and if so, the use of the native DB
function would have been selected, or would SQL database be preferred
then? The response time needs to be less the .75 second.
Thanks for advise and logic selection process.
Daniel
