Re: [alto] Potential privacy issue in draft-deng-alto-p2p-ext-01?

I'm not talking about on-demand "available" bandwidth, 

It is the bandwidth configured in the application. It is a configured value. Just take a look at any utorrent client.

A utorrent client will never use more than that amount of bandwidth, no matter your provisioned bandwidth.

________________________________________
From: Songhaibin (A) [haibin.song@huawei.com]
Sent: Friday, June 27, 2014 11:09 PM
To: Reinaldo Penno (repenno); Vijay K. Gurbani; alto@ietf.org
Subject: RE: [alto] Potential privacy issue in draft-deng-alto-p2p-ext-01?

Hi Reinaldo,

Thank you and I think you raised a very good question. I totally agree that available bandwidth is more useful than provisioned bandwidth. But available bandwidth is rather dynamic, and it is very hard to measure it and provide the real-time status to ALTO clients.

With providing provisioned bandwidth, it can be seen with a high probability a client can select a "better" peer. It is better than random IMO. But if there is an easy method to rank the available bandwidth of a peer list, I will be very interested.

BR,
-Haibin

> -----Original Message-----
> From: alto [mailto:alto-bounces@ietf.org] On Behalf Of Reinaldo Penno
> (repenno)
> Sent: Saturday, June 28, 2014 1:33 AM
> To: Vijay K. Gurbani; alto@ietf.org
> Subject: Re: [alto] Potential privacy issue in draft-deng-alto-p2p-ext-01?
>
> Another point is how¹s the provisioned access bandwidth really help decide
> which peers are better. Today most P2P software allow caps to be put for
> upload/download and people use it. Some come with a default based on the %
> of the detect access speed. So, saying a user has 1Gb/s does not really mean
> you will get better performance when connecting to him(er). I mean, it would
> more inline with better than random to get the actual bandwidth allowed.
>
>
> On 6/27/14, 10:26 AM, "Vijay K. Gurbani" <vkg@bell-labs.com> wrote:
>
> >[Still as individual.]
> >
> >On 06/26/2014 05:10 AM, Songhaibin (A) wrote:
> >> Sebastian gave an idea that we can use relative numbers to indicate
> >> the endpoint's provisioned bandwidth instead of access type, which is
> >> similar to what we have used to indicate the cost in the alto
> >> protocol.
> >
> >The difference, of course, being that the ISP in some manner consented
> >to having a normalized value of cost to be distributed in order to
> >allow for better than random selections to improve network performance.
> >
> >In the case under discussion, the issue is does the subscriber consent
> >to having their provisioned bandwidth be part of ALTO calculations?
> >
> >Remember, if the WG decides to go ahead and use provisioned bandwidth
> >anyway, it could always do so.  But then we'd better be prepared to
> >deal with the eventuality on when (and if) the IESG challenges us on
> >this privacy leak.  If that happens, we'd better have a good response.
> >
> >Perhaps a midway could be to see if we can use the provisioned
> >bandwidth for a set of (anonymous) subscribers instead of singleton
> subscribers.
> >That way, the larger herd provides some modicum of anonymity to an
> >individual subscriber who is part of the herd.
> >
> >Cheers,
> >
> >- vijay
> >--
> >Vijay K. Gurbani, Bell Laboratories, Alcatel-Lucent
> >1960 Lucent Lane, Rm. 9C-533, Naperville, Illinois 60563 (USA)
> >Email: vkg@{bell-labs.com,acm.org} / vijay.gurbani@alcatel-lucent.com
> >Web: http://ect.bell-labs.com/who/vkg/  | Calendar: http://goo.gl/x3Ogq
> >
> >_______________________________________________
> >alto mailing list
> >alto@ietf.org
> >https://www.ietf.org/mailman/listinfo/alto
>
> _______________________________________________
> alto mailing list
> alto@ietf.org
> https://www.ietf.org/mailman/listinfo/alto