1d00c920
From: barmar@alum.mit.edu
In article
<38d77dd5-99bd-4b1b-975e-45495eb137b4@a39g2000pre.googlegroups.com>,
David Schwartz wrote:
> On Dec 12, 5:09 am, karthikbalaguru
> wrote:
>
> > Also, why is the multiplication factor static ?
>
> Whatever you choose, once you've made a choice, that choice doesn't
> change.
>
> > I think, it should be dependent on various
> > other factors. Like, for example,
> > I think, it can be initially multiplied by 1.5 and
> > if it fails during that time, then the original can be multiplied
> > by 2, and if it fails during that time also, then the original
> > can be multiplied by 2.5 . Some other factors(if possible)
> > can be taken into consideration before deciding on this
> > and making the selection dynamic. Any thoughts ?
>
> Then someone will say "Why does it go 1.5, then 2, then 2.5 all the
> time? Why isn't it dynamic -- deciding what to change to based on
> other factors?".
And the answer would be that there aren't really any factors that would
tell you what to use. These numbers are all just guesses.
From what I've seen, when algorithms use exponential backoff like this,
the most common factors are either 2 or 1.5. They're probably just as
good as anything else, and easy to remember. I think the choice tends
to depend on whether you're dealing with long timeouts (like high-level
connection timeouts) or short ones (TCP retransmission timeouts). If
the initial value is something like 30 seconds, doubling it each time is
too much of an increase, so 1.5 is more common. But when you're dealing
with timeouts that start out as a fraction of a second, doubling it
isn't much.
--
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
*** PLEASE don't copy me on replies, I'll read them in the group ***
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
