[clug] broadband usage on static IP

Paul TBBle Hampson Paul.Hampson at anu.edu.au
Mon Apr 11 08:05:54 GMT 2005


On Mon, Apr 11, 2005 at 12:43:29PM +1000, Bill Clarke wrote:
> i'm wondering what the standard measure for "usage" is with a
> broadband connection with a static IP.  is it usual to charge for data
> even if the account is not logged on?

> my scenario is this: i'm a cheapskate, so i pay for the smallest
> non-shaped transact plan that suits me ($19/month, 600MB peak, 5000MB
> off-peak at velocity).  towards the end of the month i'm nearing my
> peak d/l limit, so i ensure that when i'm bit-torrenting i
> (automatically) connect after midnight and disconnect before 7am (off
> peak times). however, there are a large number of connection attempts
> from other bit-torrent users once i've disconnected.  these get
> counted by velocity as data to me and charged to my account (this may
> not sound like much, but it adds up).

> do other ISPs have this policy?

As far as Bandwidth Unlimited goes, we basically have the same policy,
although without the 7am through midnight thing, so people's IPs
shouldn't change so often that it's a problem.

Frankly, I'm not sure what a good solution _is_. The traffic does come
in to the ISP, so the ISP is having to pay for it in the end, and wants
to charge _someone_ for it.

If I could modify the network filters on the far side of my upstream
links, I'd happily script a little something that blocks traffic for
unused IPs, to prevent this sort of thing. But I doubt that's going to
happen, so I'm open to other suggestions.

Another thought that just occurred to me is to start dropping certain
destination IP/port pairs when I get 100% resets from that port, but I
fear I will trip over people's over-agressive firewalls (do you _need_
to drop incoming ICMP echo?!?) and that won't help, because we calculate
traffic at the Internet border, not the customer border (to encourage
people to Intranet/VPN/fileshare whatever with other customers freely)

Otherwise, fix every P2P protocol to _give up_ after a single RST reply,
and for Gnutella and BT, take the address off the gnodeweb's or
tracker's respective lists. I _think_ BT trackers time out a client
after a missing ANNOUNCE anyway, but I vaugely remember a warning in the
Gnutella gnodeweb stuff that says once you're on the list, you'll stay
there for an enourmous amount of time. Certainly I think most of the RST
traffic I see if Gnutella clients trying to talk to machine which is no
longer Gnutella.

-- 
-----------------------------------------------------------
Paul "TBBle" Hampson, MCSE
8th year CompSci/Asian Studies student, ANU
The Boss, Bubblesworth Pty Ltd (ABN: 51 095 284 361)
Paul.Hampson at Anu.edu.au

"No survivors? Then where do the stories come from I wonder?"
-- Capt. Jack Sparrow, "Pirates of the Caribbean"

This email is licensed to the recipient for non-commercial
use, duplication and distribution.
-----------------------------------------------------------
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.samba.org/archive/linux/attachments/20050411/fec0a23c/attachment.bin


More information about the linux mailing list