RE: RE: RE: changes to sysinfo (was Fwd: Agenda - Distributed Meeting 2010-06-23)

James,

                                                                                         

The intended scope of this discussion is much narrower I’m afraid than what you intend with the list below. I am concerned only with the definition of activeConnections as exposed through this API. As I said the determination of a preferred connection (or recommended priority) based upon QoS measures (leaving net neutrality and privacy out as a “quality” – especially since they are well beyond the scope of this discussion) is not the intent of the activeConnections attribute and should not factor into its determination.

 

Thanks, 

Bryan Sullivan | AT&T

 

From: James Salsman [mailto:jsalsman@gmail.com] 
Sent: Tuesday, June 29, 2010 2:02 PM
To: SULLIVAN, BRYAN L (ATTCINW)
Cc: James Salsman; public-device-apis@w3.org; Tran, Dzung D
Subject: Re: RE: RE: changes to sysinfo (was Fwd: Agenda - Distributed Meeting 2010-06-23)

 

Bryan,

It may be helpful to identify all aspects of service quality, because many of them are both technical and political, and we are charged with producing a set of technical policies consistent with the privacy, security, and related policies which have already been established by the group.

Then, we can ask the group which aspects of service quality should be used to determine which of the connections is active, or perhaps whether need more fine grained system information.  We have already identified some service quality aspects:

1. Able to send IP packets (as a bandwidth)

2. Able to receive IP packets (as a bandwidth)

3. Round-trip-time statistics

4. End-to-end delivery compatibility (e.g., NAT-free)

5. Network neutrality (e.g., conforming to IANA/ICANN DNS authorities without address translation)

6. Secure-compatibility (e.g., able to send HTTPS traffic without overhead)

7. Expectation of privacy (e.g. via carriers with satisfactory privacy policies and without eavesdropping)

Bryan, given the association between AT&T and eavesdropping agencies, do you believe that you should recuse yourself from discussion on that last point?

 On Jun 29, 2010 12:17 AM, "SULLIVAN, BRYAN L (ATTCINW)" <BS3131@att.com> wrote:

 James,

 A “bearer” is a transmission medium for network protocols, e.g. Ethernet, WiFi, GSM/GPRS/EDGE/3G/HSPA/LTE etc. Not all bearers are compatible with IP services (GSM SMS for example). And certainly not all bearers are equivalent in terms of QoS (in all the things that means).

  

 Indicating a preferred connection though is not the intent of the activeConnections attribute. It simply lists those connections that are compatible with IP-based services. The degree of compatibility depends upon the QoS that the application needs for the particular application.

  

 The definition of activeConnections is thus the set of connected networks (we can make it clear that these are IP networks, and that there must be an IP address assigned for the connection). 

  

 What the set of activeConnections means re selection in a multihoming environment is up to the application, or some future “connection profiling / network selection” API (this is something that we have considered in OMTP, and that factors into a variety of OMA enabler specifications, since the OMA enablers are often designed to work in multi-network environments).

  

 Like bearer preference, net neutrality is a whole other subject (or another plane of discussion at least). I would not want to bring it into this discussion, especially not do anything overt to facilitate it (or prevent it) – it’s a policy issue, not a technical issue.

 
 
  
 
 Thanks,
 
 Bryan Sullivan | AT&T
 
  

 From: James Salsman [mailto:jsalsman@gmail.com] 
 Sent: Monday, June 28, 2010 11:58 PM

 
 To: SULLIVAN, BRYAN L (ATTCINW)

 Cc: James Salsman; public-device-apis@w3.org; Tran, Dzung D
 Subject: Re: RE: changes to sysinfo (was Fwd: Agenda - Distributed Meeting 2010-06-23)

 
 
  
 
 Bryan,
 
 We need a definition of an active connection which makes sense in the context of plural...