Re: Q: Trying to understand delay pools

Pete Yandell wrote:

> So bandwidth into the bucket is throttled and the bucket has maximum size.
> But, logically speaking, what does that mean? Is the data then read by
> the client out of the bucket? Or is the bucket just a counter of sorts?

A counter that keeps track of how much data a client has used.

> How, if the "restore" parameter represents the bandwidth limit into the
> bucket, does the size of the bucket make any difference?

You can see the bucket size as a bonus system. A client not using a lot
of bandwith in average are allowed to fetch it's objects fast, but a
data hungru client who has eaten up their whole bucket size will be
strictly limited by the regain factor.

--
Henrik Nordstrom
Spare time Squid hacker