[infinispan-dev] L1 Data Container

[cotton-ben]

/> Indeed, IFF you separate the two storage areas you get in that kind of
need, but I think it's an unnecessary complexity. The world is
actually quite simple [...]/

Thanks Sanne.  Without going into any specific defense of a "complexity
sometimes needed" position, I agree with what you are saying here - both in
principle and in spirit.  Keeping it simple is always meritorious. Keeping
it simple is sound. But keeping it simple, might not be complete (at least
sometimes).

Some users do have an expectation for their provider empowering them with an
API join-point that accommodates /the user's definition/ of soundness and
completeness.  This is often viewed by the provider as a risk of adding 
"unnecessary" complexity.  This provider view is reasonable, given that
their typical users' requirement(s) cases have traditionally been
accommodated without the need of additional complexity.

Question thus becomes, is the provider willing to accommodate the
completeness requirement of  this very rare use case by introducing
complexity (complexity, which BTW,  does /nothing/ additional of value to
accommodate the completeness requirements of the provider's typical user
base).

This is where the beauty of a weighted expectation probability distribution
decision curve /may/ come into play (yet again!).  Just like some (complex
use case) L1 user wants a capability to choose "where he wants to be" on the
weighted payoff of his determining the Pr(hit at L1) vs.
Pr(readThrough at remoteCache) decision, a Cache/Grid provider wants to be
empowered with a weighted payoff decision curve that will help their
determining the probability that it is "better" to provide additional (but
rarely exposed) capability at the risk of introducing additional (and widely
exposed) complexity.

Ah, I'll be quiet now.  Thanks for letting me share my "complexity sometimes
needed for compelteness" musings.  If any interest, I can get very specific.

Ben


One last comment:  The needy "my complex use case can't be accommodated" end
user fully understands that they must  pay $ (big, and beyond the margin for
delivering additional complexity) for their completeness requirement to be
accommodated.  :-) 





--
View this message in context: http://infinispan-developer-list.980875.n3.nabble.com/infinispan-dev-L1-Data-Container-tp4027447p4027462.html
Sent from the Infinispan Developer List mailing list archive at Nabble.com.