RE: [squid-users] Delay Pools Question

From: zulkarnain <sizulku@dont-contact.us>
Date: Wed, 16 Jul 2003 01:48:13 -0700 (PDT)

--- Henrik Nordstrom <hno@squid-cache.org> wrote:
>
> What happens is that the users are constrained by
> all the constraints
> set for a pool.
>
> Global limit (class 1,2,3)
> Network limit (class 3)
> User limit (class 2,3)
>
> When multiple users are contenting for a higher
> level limit (network or
> global) Squid attempts to distribute the available
> bandwidth of that
> level more or less evenly among the users who demand
> it. As a
> consequence having significantly larger bucket sizes
> than the refill
> rate mostly makes sense on user buckets. So if you
> have a network limit
> of 30KBps and a user limit of 8Kbps then no user
> will be able to get
> more than 8KBps and each network will not be able to
> get more than
> 30KBps. If there is a demand for more than 30KBps in
> total from the same
> network then the available bandwidth (30KBps) will
> be distributed among
> the users who demand the bandwidth.
>

Based on these info, I'm trying to convert class 2
delay pools to class 3 delay pools to meet above
requirements;

--Begin class 2 delay pools-->

delay_pools 1
delay_class 1 2
# 384 kbps max for the group, 10kbps for each user.
delay_parameters 1 48000/48000 1250/1250
# ACLs to apply to the delay pools
acl lowspeed src 192.168.1.1-192.168.1.192
delay_access 1 allow lowspeed
delay_access 1 deny all

<--End--

The changes are;

--Begin class 3 delay pools -->

delay_pools 1
delay_class 1 3
# 384 kbps max for the group, 10kbps for each user.
delay_parameters 1 48000/48000 48000/48000 1250/48000
# ACLs to apply to the delay pools
acl lowspeed src 192.168.1.1-192.168.1.192
delay_access 1 allow lowspeed
delay_access 1 deny all

<--End--

is it correct?

regards,
sizulku

__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com
Received on Wed Jul 16 2003 - 02:48:18 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:18:01 MST