RE: [802.21] Tomorrow's Telecon material
Hi Vivek,
I agree that proactive info queries are necessary for timely decisions
which is certainly implementation dependent. But the nature of IS
queries may involve (may be only L3) proxies, multiple transport
connections and database queries. Putting a requirement for time
sensitivity seems like a good idea to me. As you say this feature can be
used/misused but I don't see downside in using it either. The intention
is to ensure that the transport design allows for quick queries.
Regards,
Srini
>-----Original Message-----
>From: ext Gupta, Vivek G [mailto:vivek.g.gupta@intel.com]
>Sent: Friday, March 17, 2006 8:29 AM
>To: Sreemanthula Srinivas (Nokia-NRC/Dallas);
>Qiaobing.Xie@MOTOROLA.COM; STDS-802-21@listserv.ieee.org
>Subject: RE: [802.21] Tomorrow's Telecon material
>
>>
>>The need for timely delivery is more context based than it is content
>based. >I mean the urgency is known to only to the requesting
>entity based on its >current conditions and therefore, setting
>a delay requirement for transport >does make sense to me.
>>
>
>It might help to understand this client context better,
>why/how did the client get into such a state...before getting
>too far into the solution space. For all you know the reason
>for this client state (context) could just be a not so good
>implementation as opposed to a more germane design requirement.
>If clients don't get information in a proactive manner and
>plan for vertical handovers..., any amount of delay
>sensitivity built into transport may not be good enough to
>rescue the connection under adverse situations. Also even if
>the use cases and requirements are quite relevant, just a
>delay sensitive transport may not be good enough to rescue
>such cases....
>
>That said, in general the transport should work well for most
>normal cases. If there is a delay sensitive option in
>transport that comes for free I can easily see clients abusing
>it and using it as a default anyway....
>
>Best Regards
>-Vivek
>
>> -----Original Message-----
>> From: stds-802-21@ieee.org [mailto:stds-802-21@ieee.org] On
>Behalf Of
>> Srinivas.Sreemanthula@NOKIA.COM
>> Sent: Thursday, March 16, 2006 3:46 PM
>> To: Qiaobing.Xie@MOTOROLA.COM; STDS-802-21@listserv.ieee.org
>> Subject: RE: [802.21] Tomorrow's Telecon material
>>
>> Qiaobing,
>> Inline ...
>>
>> Regards,
>> Srini
>>
>> >-----Original Message-----
>> >From: ext Qiaobing Xie [mailto:Qiaobing.Xie@MOTOROLA.COM]
>> >Sent: Thursday, March 16, 2006 3:51 PM
>> >To: STDS-802-21@listserv.ieee.org
>> >Subject: Re: [802.21] Tomorrow's Telecon material
>> >
>> >Srini,
>> >
>> >Sorry for missing your earlier question. But if only some
>> >transactions are time sensitive, would setting a general delay
>> >requirement for transport (i.e., forcing every transaction
>to pay the
>> >same cost) become a little wasteful?
>>
>> Srini>Good point, not all IS queries need it. But for some instances
>> that need it, the transport must support it. I think the transport
>must
>> provide this capability and also ability to switch off the
>feature, if
>> there is no need.
>>
>> >Moreover, IP in general is best effort network and every
>transport is
>> >at the mercey of the underneath IP layer. I don't know any existing
>> >mechanism that allows the sender to control the timeliness of
>> >delivering a packet over an IP path (the most a sender can do is to
>> >set the TOS bits and pray).
>>
>> >The more common practice for a sender is to set a shorter
>timer if it
>> >knows that an out-going message is urgent so that it can engage
>> >failure handling sooner if the expected response does not come back
>> >from the far-end.
>> >
>> >By the way, the only time sensitive IP transport defined in IETF is
>> >RTP, while neither TCP or UDP is considered time sensitive.
>>
>> Srini> I am not sure how this requirement will be supported. One way
>> would be use datagram services that don't need session setup.
>>
>> >
>> >regards,
>> >-Qiaobing
>> >
>> >Srinivas Sreemanthula wrote:
>> >
>> >> Hi all,
>> >> I am answering to my own email. The need for timely delivery is
>more
>> >> context based than it is content based. I mean the urgency
>> >is known to
>> >> only to the requesting entity based on its current conditions and
>> >> therefore, setting a delay requirement for transport does make
>sense
>> >> to me.
>> >>
>> >> Regards,
>> >> Srini
>> >>
>> >>
>> >>>-----Original Message-----
>> >>>From: ext Srinivas Sreemanthula
>> >>>[mailto:Srinivas.Sreemanthula@nokia.com]
>> >>>Sent: Saturday, March 11, 2006 8:18 AM
>> >>>To: STDS-802-21@listserv.ieee.org
>> >>>Subject: Re: [802.21] Tomorrow's Telecon material
>> >>>
>> >>>Qiaobing,
>> >>>Without taking the discussion into classifying static or dynamic
>> >>>information... are there any time-sensitive information
>in our IE,
>> >>>now?
>> >>>
>> >>>Regards,
>> >>>Srini
>> >>>
>> >>>
>> >>>>-----Original Message-----
>> >>>>From: ext Qiaobing Xie [mailto:Qiaobing.Xie@motorola.com]
>> >>>>Sent: Friday, March 10, 2006 6:09 PM
>> >>>>To: Sreemanthula Srinivas (Nokia-NRC/Dallas)
>> >>>>Cc: STDS-802-21@listserv.ieee.org
>> >>>>Subject: Re: [802.21] Tomorrow's Telecon material
>> >>>>
>> >>>>....
>> >>>>
>> >>>>>whether there is need for "timely delivery of IS" or not.
>> >>>>
>> >>>>The answer depends on what content the IS is carrying. If
>> >the content
>> >>>>is time sensitive, "yes", otherwise, "no".
>> >>>>
>> >>>>regards,
>> >>>>-Qiaobing
>> >>>>
>> >>>>
>> >>>
>> >>
>> >
>