Just a note for everybody interested in the group/community
management topic:
Due to the ongoing IETF meeting and the peer-protocol design discussions,
there is a lot of activity on the P2PSIP mailing list.
Thus, it is probably wise to postpone the group management discussion
a little (at least on this mailing list) so that the topic won't get
"buried" under the other topics.
Regards,
- Otso
P.S. Unfortunately I was not able to take part in IETF 71.
On Thu, 28 Feb 2008, Otso Kassinen wrote:
Hi Victor and Radhika.
What do you have in mind?
We should first determine, which is the better approach:
"a group is an overlay" or "many groups per overlay",
just as you pointed out. It must be studied, what can be done
with the communities in these approaches; is there something that
(for example) restricts group functionality if plain overlays are used.
Group management is often an integral part of P2P environments,
especially in person-to-person communication systems.
This kind of protocols can be studied to get ideas for our solution.
is there any significant difference between (SIP) (sub)domains
concept and communities?
I think, groups are more dynamic than a node just belonging to a specific
domain, and overlapping of groups is easier (but the concepts are similar).
The "server farm style" concept for reliability and load sharing
(mentioned by Radhika) seems to emphasize the redundancy of nodes.
That is clearly related to P2PSIP. However, we are primarily thinking of
group management where non-redundant nodes - a node per user - are taking
part in some activity or interest, person to person (e.g. games, IM).
Gallup / questionnaire for those interested in the Community topic:
* Does anyone have an idea, what existing WG could be suitable for taking
the Community idea further?
* Alternately, if there's no suitable WG, we could consider forming
a new WG. This would be quite a big step; if a new WG is started,
there should be some broader agenda than just P2PSIP group management.
For example, Radhika's idea about peer robustness could be included?
Best regards,
- Otso
On Wed, 27 Feb 2008, Roy, Radhika R Dr CTR USA USAMC wrote:
Hi, Victor:
It is an interesting concept: Community.
We like to see this paper.
Best regards,
Radhika
PS: We used to think like server farms. A server farm many have many servers
for providing reliability and load sharing purposes. However, the logical
view of the server has been a single server using standardized protocol(s)
to the outside world. The communication within the servers was proprietary.
In the case of "community" concept, I assume, that we can use the similar
concept like that of the server farm. However, there is a difference. Now,
the communications protocols used within a community will be standardized,
as I guess you will be proposing in your draft.
It will extend the P2PSIP concept for making a peer more robust - as I see
it - a "logical" P2PSIP peer will probably NEVER fail because it is a part
of a larger community.
Definitely, it is welcomed to see the concept. If it is needed, we can form
another WG to standadize this "community" P2PSIP protocol.
----- Original Message -----
From: Victor Pascual Ávila
Date: Wednesday, February 27, 2008 13:10
Subject: Re: [P2PSIP] Group management on top of P2PSIP?
To: Otso Kassinen
Cc: [email protected]
Otso,
On Wed, Feb 27, 2008 at 4:31 PM, Otso Kassinen
wrote:
Has anyone already come up with some ideas or even an
implementation> of a decentralized group-management system on top
of P2PSIP?
(I searched through the P2PSIP mailing list archives
and found no mention of group management.)
Some time ago, we (Pompeu Fabra University) considered including
'Community' concept in the P2PSIP terminology.
We can think about 'Community' as a group of nodes (Overlay layer) or
as a group of users (upper application layer, e.g. SIP).
What do you have in mind?
On the one hand, we considered community/group/whatever as a group of
nodes and our main discussion was about if using a single overlay for
both communities or just using one overlay (routing, storage) for each
community. On the other hand, if we consider community as a group of
users on the top of a single (or many) overlays, is there any
significant difference between (SIP) (sub)domains concept and
communities? I'm not trying to group users into domains, just
suggesting a similar concept.
We are preparing a paper on a related topic. It's basically a P2PSIP
live streaming tv platform where users are grouped into
stations/channels. Definitely, it is not in the scope of the P2PSIP
WG.
Cheers,
--
Victor Pascual Ávila
Research Engineer
Tel. +34 93 542 2906
Fax. +34 93 542 2517
Research Group on Network Technologies and Strategies (NeTS)
Universitat Pompeu Fabra (UPF)
Pg. de Circumval·lació, 8
Office 358
08003 Barcelona (Spain)
http://nets.upf.edu/
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip
