RE: [p2pi] WG Review: Application-Layer Traffic Optimization (alto)

Hi Vijay,
I am not at all talking about reinventing what BitTorrent can do or even 
remotely about any actual p2p application itself.  I am only talking about peer 
selection.  However, I think there is a critical difference between what I view 
as contributing to peer selection and what the current proposed charter does.

Peer selection is important to ISPs from a network utilization perspective and 
to peers themselves from a performance perspective.  That automatically makes 
peer selection a function of multiple aspects - a) information that some 
service providers may decide to share with the peers, b) information that peers 
decide to make available about themselves to other peers for this purpose, and, 
c) any measurements peers may do on their own.  The current charter definition 
(and from what I can tell based on your response below) only seems to allow for 
a).  I would agree that c) is out of scope of ALTO and something that peers can 
additionally do.  I strongly believe that b) should be part of the ALTO work.  
This functionality itself is application agnostic and requires an interoperable 
solution for it to be beneficial.  This has nothing to do with content itself; 
it is purely about sharing information that helps with peer selection.

Lastly, as long as this framework is made available, information can be shared 
among peers on an overlay in a distributed fashion and/or provided by a service 
provider host.

Best regards,
Vidya

-----Original Message-----
From: Vijay K. Gurbani [mailto:vkg(_at_)alcatel-lucent(_dot_)com]
Sent: Monday, October 13, 2008 8:50 AM
To: Narayanan, Vidya
Cc: IESG IESG; ietf(_at_)ietf(_dot_)org; p2pi(_at_)ietf(_dot_)org
Subject: Re: [p2pi] WG Review: Application-Layer Traffic
Optimization (alto)

Vidya: Thank you for your response and your time in helping
define the work.  More inline.

Narayanan, Vidya wrote:

When we consider ALTO as a distributed service, there may not
necessarily be "a" host that specifically resolves the ALTO queries.
 For instance, consider the case where ALTO is a service

offered in an

overlay.  There may be peers publishing information about

themselves

on the overlay and other peers looking up such information.

 These are

not necessarily client-server style communications.  In

fact, all that

is important in this context is that the overlay acts as a

rendezvous

for sharing such information.


I think the disconnect we may be having is that you view ALTO
as a peer description protocol; it is not.  Other protocols
like BitTorrent, for example, are more suited to this, and
they do exactly what you want.  In a BitTorrent overlay
(swarm), the overlay knows exactly which peer is contributing
which content, which peer has which chunks, the
download/upload ratio, the time the peer joined the swarm,
whether the peer is choked or unchoked, whether the peer has
a public port, etc.  ALTO is not out to replace BitTorrent.
What ALTO is providing are better strategies for peer selection.

For instance, it is not ALTO that gets to decide which peer
is hosting which content and what the contributions of that
peer to the overlay are.  However, it is ALTO's job to
provide information to a querying peer allowing it to
determine wisely where it will download the content from.

I'm afraid that would be a mistake.  It actually doesn't

matter if we

don't agree today on the exact types of information that can be
shared.  It is important that we have a protocol that

allows peers to

publish ALTO related information.  Having this protocol be

extensible

would allow for any type of information to be carried in it.


So far, no one on the list has proposed that ALTO be a peer
description and publication protocol.  So based on the
discussion we have had since (essentially the workshop in)
May 2008 on the p2pi list, I would hesitate to add in the
charter something that participants have not expressed any
preference for (i.e., a deliverable on peers publishing their
information.)

Actually, I am saying that is exactly what is not needed.

I don't see

the information types as something this effort will

necessarily nail

down.


I am confused; I thought earlier you were trying to make the
case that ALTO should provide even more specific information
that needs to be published?

In the end, we do agree on that any protocol be extensible.
Whether that is extensible through a registry-like mechanism
or other means remains to be discussed in the WG, right?

I would like us to think beyond applications we see today.

Had TCP not

been designed that way, we probably would have needed a

redesign of it

for HTTP :)


Protocols evolve, networks evolve.  I am sure we were
prescient when we designed TCP such that HTTP could simply
use it.  However, other realities of evolution did force us
to design SCTP, for instance.  But regardless, the point is
that we should be general, and we are; but we have to do this
while drawing a fine line between research and engineering.
Some applications that are appearing on the horizon are
streaming media and P2PTV; for these we have opened up
channels with IRTF.  So I don't see where we are constraining
ourselves in ALTO.

I can envision video applications using the P2PSIP framework, for
instance.


Yes, and as I wrote earlier, they can use ALTO to discover
the the peer they find optimal within their constraints and use it.

In any event, I still don't have a good understanding of

what it means

to consider the needs of these various things - what does

it mean to

say that we'll consider the needs of BitTorrent/CDN, etc.?
Could you maybe give me an example of what it means?


Look at the Ono work, which is a plug-in to BitTorrent.  It
uses Akamai redirections to find the closest peer to download
content from.  In a sense, ALTO is replacing that ad-hoc
lookup and providing a much more deterministic answer.  That
is what we mean by the needs of "BitTorrent/CDN etc."

What I am saying is that it is not for us to determine the

usefulness

of a particular piece of information.  As long as the peers

or service

providers are willing to share a piece of information, that can be
consumed by other peers as they deem fit.  So, I don't

think we should

consider ourselves the gatekeeper for the types of

information shared.

But we are not.  As I made the point earlier, ALTO is not out
to replace BitTorrent.  So replicating in ALTO the details
about peers that BitTorrent already has is
counter-productive.  Instead, ALTO can focus on providing the
pieces that BitTorrent does not:
topology, policy, etc.  That is where we will make a difference.

Thanks,

- vijay
--
Vijay K. Gurbani, Bell Laboratories, Alcatel-Lucent 1960
Lucent Lane, Rm. 9C-533, Naperville, Illinois 60566 (USA)
Email: vkg(_at_){alcatel-lucent(_dot_)com,bell-labs.com,acm.org}
WWW:   http://www.alcatel-lucent.com/bell-labs

_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf