Some comments inline.

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Song 
Haibin
Sent: Tuesday, February 10, 2009 6:22 PM
To: 'Eric Burger'
Cc: [email protected]
Subject: Re: [alto] differences among applications

I don't like the idea to standardize the peer selection algorithms for each
application either. 

But image that when ALTO server receives a request with a source peer and a
destination peer list for peer selection, it responds the ranking
information for each peer in the list. Here with different applications the
ranking result may be different. I don't know if ALTO server must implement
different peer selection algorithms internally for different applications,
so that it can rank the candidates list.

>>To enable this we need a way for an ALTO service to specify the metrics it 
>>can provide guidance on and a way for the application to select a metric of 
>>interest. Similar to P2PSIP, ALTO could have a alto service configuration 
>>document that provides such information. The p4p alto service could say that 
>>it supports the pDistance metric with a rating scale from 1 to X. Another 
>>alto service could use a different metric or multiple metrics. The 
>>application should have a way to specify the metric of interest. The ALTO 
>>service can then implement the peer selection which is just a ranking 
>>according to a metric it supports. It need not infer the application 
>>requirements.

Or ALTO server does not need to understand the requirement of a specific
application, surely, it can responds to queries with parameters, e.g. the
ALTO server will not responds with the ranking info (generally a simple
ranking info will take a lot of considerations into account), but just the
topology information such as the routing hops from peer A to peer B is 5,
peer A to peer C is 4, the link quality from peer A to peer B is "good",
peer A to peer C is "fair", the economy cost..... And then with retrieval
many types of information what the application needs, the application makes
the choice.

>>I think the para above is a good direction. However, it is not clear how to 
>>tag the ranking info with good enough information to allow the application to 
>>make a good selection. The ALTO service potentially should be able to cull 
>>the list of not useful peers if some of them do not satisfy some constraint 
>>in the query and not have to return all annotated ranking info back to the 
>>client.

So my general confusion here is whether the ALTO server needs to understand
the requirements of a specific application to do the peer selection.

>>by requirement of a application if you mean the fact that an application can 
>>query for a metric of interest or specify a threshold (e.g. >128kbps), the 
>>ALTO server should likely be able to do this. It should not need to however 
>>infer the intent of the querying application i.e. giving a peer selection 
>>good for a p2p streaming application or good for a download service. The 
>>application should have to translate its needs to a standardized query 
>>supported by the ALTO server.

Best
Saumitra

Best Regards,
Haibin
Email: [email protected]
Skype: alexsonghw

>-----Original Message-----
>From: Eric Burger [mailto:[email protected]]
>Sent: Tuesday, February 10, 2009 9:53 PM
>To: Song Haibin
>Cc: [email protected]
>Subject: Re: [alto] differences among applications
>
>I would offer it is important to have the client specify what it
>thinks is important. However, I would also offer it would be fatal to
>have "Profile B", "Profile N", "Profile S" selection algorithms, where
>B, N, and S are different applications. I will guarantee that by the
>time we're done in the IETF, no one will case about those applications
>and will have moved on to some other, hot applications.
>
>It may be worth noting what parameters are important.
>
>On Feb 10, 2009, at 7:25 AM, stefano previdi wrote:
>
>>
>> On Feb 10, 2009, at 12:43 PM, Song Haibin wrote:
>>
>>> I think it is necessary to discuss whether we need to standardize
>>> different
>>> peer selection algorithms according to different types of
>>> applications.
>>
>> We may want the alto protocol to allow the requester to specify
>> which type of ranking/preference it needs. Note that this doesn't
>> mean we need to standardize any algorithm.
>>
>> s.
>>
>>>
>>> Best Regards,
>>> Haibin
>>> Email: [email protected]
>>> Skype: alexsonghw
>>>
>>>
>>>> -----Original Message-----
>>>> From: [email protected] [mailto:[email protected]] On
>>>> Behalf Of
>>> Enrico
>>>> Marocco
>>>> Sent: Tuesday, February 10, 2009 7:16 PM
>>>> To: Zoran Despotovic
>>>> Cc: [email protected]
>>>> Subject: Re: [alto] differences among applications
>>>>
>>>> Zoran Despotovic wrote:
>>>>> I was wondering if and how IETF would address possible
>>>>> differences among
>>>>> relevant P2P applications in the sense that different
>>>>> applications may
>>>>> require totally different solutions. Was there any discussion on
>>>>> this
>>>>> before on the list?
>>>>
>>>> Yes, there's been some discussion about how to deal with the fact
>>>> that
>>>> different applications may have different requirements for peer
>>>> selection -- e.g. file-sharing would benefit most from connections
>>>> to
>>>> peers with higher uplink bandwidth, while realtime communications
>>>> applications would probably want to chose media relays with the
>>>> smallest
>>>> delay.
>>>>
>>>> I remember someone suggested that a way to address it could be to
>>>> simply
>>>> provide a means for the querying peer to specify what parameters it
>>>> would like to have optimized (delay, bandwidth...) and let the ALTO
>>>> server use such information in its processing. However, AFAIK, no
>>>> solution proposals at this time do anything like that.
>>>>
>>>> --
>>>> Ciao,
>>>> Enrico
>>>
>>> _______________________________________________
>>> alto mailing list
>>> [email protected]
>>> https://www.ietf.org/mailman/listinfo/alto
>>
>> _______________________________________________
>> alto mailing list
>> [email protected]
>> https://www.ietf.org/mailman/listinfo/alto


_______________________________________________
alto mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/alto
_______________________________________________
alto mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/alto

Reply via email to