RE: [P2PSIP] do we need a client protocol?

[<[e-mail]>]

Some comments inline...


________________________________

	From: ext HUANG Ping B [mailto:[e-mail]] 
	Sent: 12 July, 2007 11:57
	To: Zheng Hewen; Michal Giermasinski; [e-mail]
	Subject: RE: [P2PSIP] do we need a client protocol?
	
	
	I have never said that P2PSIP is equal to SIP, P2PSIP is just an overlay as an alternative to the SIP discovery process of RFC 3263, P2PSIP offers an alternative mechanism for determining the correct destination for SIP requests. 
	 
	Do you need SIP to offer an alternative mechanism for determining a correct destination for SIP requests?
	 
	SIP is session layer protocol, P2PSIP is under SIP. of cause, is you use another protocol, P2PSIP may be called another name, such as P2PXXX. Peers in P2PSIP overlay are responsible not just for location (done by P2PSIP protocol). Why we called P2PSIP as "P2PSIP"?  this name means that there is a relationship between P2PSIP and SIP, maybe SIP over P2P or P2P over SIP. but any way, SIP should be used. 
	 
	SIP is a session initiation protocol. In some cases it may be used only between SIP UAs.
	 
	  then Peers in P2PSIP overlay( not P2PXXX overlay) should recognize SIP message. Ok, you can say that a tunnel protocol could be used to deliver SIP protocol, this is a choice, but not good choice. if a tunnel protocol is used, then SIP has nothing with P2P overlay, this tunnel protocol could deliver any protocols besides SIP, then name "P2PSIP" does not make sense.
	when peers in P2PSIP overlay do not recognize SIP message, then let them out of overlay. 
	 
	I would like to encourage you to read the following drafts:
	http://www.p2psip.org/drafts/draft-bryan-p2psip-requirements-00.txt
	http://www.ietf.org/internet-drafts/draft-baset-p2psip-p2pp-00.txt <http://www.p2psip.org/drafts/draft-baset-p2psip-p2pcommon-01.txt> 
	http://www.p2psip.org/drafts/draft-jennings-p2psip-asp-00.txt 
	 
	Marcin
	 

________________________________

	From: Zheng Hewen [mailto:[e-mail]] 
	Sent: Thursday, July 12, 2007 2:01 PM
	To: HUANG Ping B; 'Michal Giermasinski'; [e-mail]
	Subject: RE: [P2PSIP] do we need a client protocol?
	
	
	Hi,
	 
	  If I do not misunderstand, P2PSIP is not equal to SIP.
	  Essentially in P2PSIP overlay, P2PSIP peers have nothing with SIP. Why must we bind SIP to P2PSIP peers? 
	  
	  Here is the description about P2PSIP peers in WG charter, FYI.
	  "The work focuses on collections of nodes called "P2PSIP peers" and "P2PSIP clients". P2PSIP peers manifest a distributed namespace in which overlay users are identified and provides mechanisms for locating users or resources within the P2PSIP overlay. P2PSIP clients differ from P2PSIP peers primarily in that they do not store information in the overlay, but only use it to locate users and resources. P2PSIP clients and peers use the resolution services of the peers as an alternative to the SIP discovery process of RFC 3263. In this way, P2PSIP offers an alternative mechanism for determining the correct destination for SIP requests. The working group's initial charter scope will be to produce protocols to enable this alternate mechanism for RFC 3263 functionality."
	 
	  Nothing says that P2PSIP peers must support SIP or have anything with SIP.
	  From the charter, functionality of P2PSIP peers is so clear and simple, why do we make it complicated? I think we can do it simpler, SIP is SIP, P2PSIP peer is P2PSIP peer, let them be.
	 

	

	

	Zheng Hewen
	HUAWEI TECHNOLOGIES CO.,LTD.   

	


	Huihong Mansion
	No.91 Baixia Road
	NanJing, P.R.China
	Tel: [phone]
	Fax: [phone]
	Mobile: [phone]
	E-mail: [e-mail]
	www.huawei.com
	-------------------------------------------------------------------------------------------------------------------------------------
	This e-mail and its attachments contain confidential information from HUAWEI, which 
	is intended only for the person or entity whose address is listed above. Any use of the 
	information contained herein in any way (including, but not limited to, total or partial 
	disclosure, reproduction, or dissemination) by persons other than the intended 
	recipient(s) is prohibited. If you receive this e-mail in error, please notify the sender by 
	phone or email immediately and delete it!
	
	 

________________________________

	From: HUANG Ping B [mailto:[e-mail]] 
	Sent: 2007年7月12日 10:22
	To: Zheng Hewen; Michal Giermasinski; [e-mail]
	Subject: RE: [P2PSIP] do we need a client protocol?
	
	
	Hi all
	    P2P is great topic, but P2PSIP is the subset or certain instance of P2P. when we discuss P2PSIP, we shall not involve all nodes which could be peers in P2PSIP overlay. peers in P2PSIP overlay are not equal to peers in P2P overlay. when we discuss IMS, although legacy equipments does not support SIP, but some AGWs do translation or agent for those equipments. then, when we discuss P2PSIP, shall we involve all peers who know SIP or not in P2PSIP overlay? I think P2P is not equal to P2PSIP, and vice versa. and I think that Peers in P2PSIP overlay should have some common base functionalitis such as SIP recognition. I think we shall make P2PSIP more simple but not more complex.
	
________________________________

	From: Zheng Hewen [mailto:[e-mail]] 
	Sent: Thursday, July 12, 2007 9:34 AM
	To: 'Michal Giermasinski'; [e-mail]
	Subject: RE: [P2PSIP] do we need a client protocol?
	
	
	Hi,
	 
	    Please let me describe the case in detail.
	    In the case, all clients support SIP, but some peers in the overlay network do not support SIP and some support SIP. For those peers which do not support SIP, such as some PCs, they can provide all required DHT services, but there is only a pity that they can not recognize any SIP message, maybe they think that the capability is unnecessary. In this case, some client protocol is desirable. Certainly this protocol may be the subset of general Peer Protocol if we regard those clients as Lame Peers and regard those peers as Pure Peers who only provide required DHT services.
	    IMHO, this case does not care those clients not supporting SIP.
	 
	Zheng Hewen
	HUAWEI TECHNOLOGIES CO.,LTD.   

	


	Huihong Mansion
	No.91 Baixia Road
	NanJing, P.R.China
	Tel: [phone]
	Fax: [phone]
	Mobile: [phone]
	E-mail: [e-mail]
	www.huawei.com
	-------------------------------------------------------------------------------------------------------------------------------------
	This e-mail and its attachments contain confidential information from HUAWEI, which 
	is intended only for the person or entity whose address is listed above. Any use of the 
	information contained herein in any way (including, but not limited to, total or partial 
	disclosure, reproduction, or dissemination) by persons other than the intended 
	recipient(s) is prohibited. If you receive this e-mail in error, please notify the sender by 
	phone or email immediately and delete it!
	
	 

________________________________

	From: Michal Giermasinski [mailto:[e-mail]] 
	Sent: 2007年7月12日 5:31
	To: Zheng Hewen; [e-mail]
	Subject: Re: [P2PSIP] do we need a client protocol?
	
	
	Hi!
	 
	Well, in my oppinion it is not a good for creating a new protocol. If a client application is not able to use SIP protocol that is well known for years it obviously will not be able to use a new one that will be designed especially for cooperation with p2psip overlay. 
	 
	If we follow this understanding we can assume that the better thing is to use SIP on this interface (between client outstanding the overlay and the peer in the overlay). The only issue is to create a proxy-like peer that will be able to help client to get benefits of the overlay. 
	 
	What we should consider is how to filter the traffic unwanted by regular client that do not want to take part in the routing decision.
	 
	best regards
	Michal Giermasinski


	 
	2007/7/11, Zheng Hewen <[e-mail]>: 

		Hi,
		
		   A client protocol will be useful when some peers in the overlay do not support SIP, those peers are only responsible to 
		provide distributed storage, location and transport services for information about SIP, but those peers do not recognize any
		SIP messages at all.
		  In this case, how to maintain the relation between clients (UA) and the hosting peer(s)? One client protocol is desirable 
		at this time.
		
		
		Zheng Hewen
		HUAWEI TECHNOLOGIES CO.,LTD.  huawei_logo
		
		
		Huihong Mansion
		No.91 Baixia Road
		NanJing, P.R.China
		Tel: [phone]
		Fax: [phone]
		Mobile: [phone]
		E-mail: [e-mail]
		www.huawei.com
		---------------------------------------------------------------------------------------------------------------------------- 
		---------
		This e-mail and its attachments contain confidential information from HUAWEI, which
		is intended only for the person or entity whose address is listed above. Any use of the
		information contained herein in any way (including, but not limited to, total or partial 
		disclosure, reproduction, or dissemination) by persons other than the intended
		recipient(s) is prohibited. If you receive this e-mail in error, please notify the sender by
		phone or email immediately and delete it! 
		
		
		-----Original Message-----
		From: Henry Sinnreich [mailto:[e-mail]]
		Sent: 2007?6?26? 1:39
		To: Dean Willis; David A. Bryan
		Cc: Avshalom Houri; P2PSIP Mailing List; Philip Matthews 
		Subject: [P2PSIP] do we need a client protocol?
		
		Folks,
		
		Given the length of the discussion on "do we need a client protocol?"
		let me repeat the simple imperatives for the client:
		
		- Limited battery in mobile devices, and 
		
		- Other uses besides SIP. Business planning folks may not care about the WG charter, but about the return on the investment.
		
		Thanks, Henry
		
		-----Original Message-----
		From: Dean Willis [mailto: [e-mail] <mailto:[e-mail]> ]
		Sent: Thursday, June 14, 2007 11:23 PM
		To: David A. Bryan
		Cc: Avshalom Houri; P2PSIP Mailing List; Philip Matthews
		Subject: Re: [P2PSIP] Concepts document: proposal for active/passive peerterminology 
		
		David A. Bryan wrote:
		> If I understand this right, not sure I like it. Do these
		"non-contributing
		> peer" (I *do* like contributing better than active/passive) entities
		have a
		> Peer-ID? In otherwords, are they part of the overlay at all? If so, 
		for
		> most
		> DHTs, how do we allow you to be in the overlay and not store resources
		(not
		> typical for most current DHT algorithms)
		
		Well, that certainly depends on the DHT algorithm involved. It certainly seems that a peer that moves between contributing 
		and non-contributing might HAVE a peer ID, but whether it chooses to insert itself into the DHT "right now" might determine
		whether it is contributing "right now"
		or not.
		
		> My concern all along has been that I don't understand why something is 
		> a peer if it isn't doing anything other than talking to someone and
		> making requests, hence why I supported client in the first place.
		>  I fail to see how something that isn't contributing anything, and 
		> instead talks to a peer to get resources, is in any way a peer. It
		> seems completely not in keeping with the idea of P2P. I also worry it
		> is likely to be very confusing since we've been using the other 
		> terminology for over a year now. Just looks like a souce of confusion.
		
		A node might sometimes have a high-bandwidth connection and wish to contribute resources, and sometimes not have a
		high-bandwidth connection and therefore not wish to offer resources. Does this modality make it two different nodes? 
		
		> If we do go this way, there is one subtle point. If I join the overlay
		but
		> don't happen to offer any services (other than maybe
		storage...depending on
		> how we definie it, that might or might not be mandatory), would I be 
		> contributing? If joining the overlay (in the sense that we have a
		PeerID
		> and
		> we route traffic) doesn't require me to store, I could offer no
		services
		
		That's "offer no other services", since routing is itself a service. 
		
		> and
		> still be a peer. How do we distinguish between this case and a
		> "non-contributing peer" who isn't even part of the overlay (i.e., no
		> Peer-ID)
		
		I think there's a range of "contributing" in play. For most DHT models, a minimally contributing node will route a request 
		it receives. It might ask nicely (perhaps a q-value mechanism) to not have to route requests, but would probably
		mathematically still have to route something it gets.
		
		I think we distinguish based on the service advertisements from the peer. 
		
		> If we are so opposed to client for confusion with SIP UAs, lets pick a
		new
		> name, but why some flavor of peer when it does nothing remotely peer
		like?
		
		Because in 15 minutes it might roam onto a WiFi net and therefore change it service offering. 
		
		--
		Dean
		
		_______________________________________________
		P2PSIP mailing list
		[e-mail]
		https://www1.ietf.org/mailman/listinfo/p2psip 
		
		_______________________________________________
		P2PSIP mailing list
		[e-mail]
		https://www1.ietf.org/mailman/listinfo/p2psip 
		
		
		
		
		_______________________________________________
		P2PSIP mailing list
		[e-mail]
		https://www1.ietf.org/mailman/listinfo/p2psip 
		

_______________________________________________
P2PSIP mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/p2psip