Re: Greylisting BCP

> -----Original Message-----
> From: Jose-Marcio Martins da Cruz [mailto:[email protected]]
> Sent: Wednesday, October 26, 2011 2:47 AM
> To: Anti-Spam Research Group - IRTF
> Cc: Murray S. Kucherawy
> Subject: Re: [Asrg] Greylisting BCP
> 
> You should take some care with this. Ressource usage varies a lot with different implementations :
> you'll find in memory chained lists, some RDBMS (e.g. MySQL) or ISAM/B-Tree/HASH (e.g. BerkeleyDB).
> Some implementations are smarter than others about ressources needes :
> e.g. how to expire old entries or how to aggregate them. And this makes
> a big difference on how expensive the filter is.
> 
> Put simply, when talking about greylisting, expensive or not is more
> related to the implementation and to some tuning (e.g. delays) than to
> the method itself. I'm not even talking about poor programming
> practices...

The point is that there is a range of expenses based on the implementation, the cost is rarely zero or even close to it.  I don't think it's necessary to run up a full bill of materials for each possible implementation in the BCP, but merely to make it clear that the benefits have a definite cost associated with them in terms of resources.