[cap-talk] Price of resource accountability

[Neal H. Walfield]

At Fri, 05 Sep 2008 09:47:20 +1000,
James A. Donald wrote:
> 
> Neal H. Walfield wrote:
> > This question makes the assumption that resource accountability is
> > primarily useful for limiting denial of service attacks.  I think this
> > is a weak justification for resource accountability.  A stronger
> > justification, I think, is that good resource accountability means
> > that it is possible to more accurately enforce resource distribution
> > policies, and provide information to resource principals about the
> > amount of resources available to them.  This is one of the main
> > observations of the Viengoos work.
> 
> Google's new browser is supposed to report on which web pages are 
> sucking up unreasonable amounts of resources.  I often have a very large 
> number of web pages open at once, making it difficult to determine which 
> web page is bringing my system to a grinding halt.
> 
> Having determined what web site is responsible, I would promptly 
> blacklist it.

That's a variation of the denial of service attack.  (And, not a very
good one: sometimes backlisting is too extreme.)

What I am interested in is determining how to allow the stakeholders
to control how resources are distributed among their child
computations.  This becomes increasingly important when trying to
answer the question of how much resources adaptive applications should
use.  If you are interested in more details both about the problem and
my solution, feel free to take a look at a draft version of my thesis:

   http://plato.walfield.org/20080729-walfield-viengoos-thesis-final.pdf

Neal