On Fri, Apr 29, 2011 at 4:57 PM, stefano previdi <[email protected]> wrote:
> On Apr 29, 2011, at 3:12 AM, 朱潇 wrote: > >> hi Sebastian ： >> * firstly, i agree that the creation of the map, which is tightly >> related to the algorithm, is within the protocol. >> > > > the protocol specifies the semantic of the map content. How you > came to that (from whatever source) is out of protocol spec. > > > * secondly, i am not sure the information contained in current map >> structure is complete or not, because the current protocol describe that : >> >> * the cost map just includes cost type and cost mode >> > > > I believe the cost mode will reflect the topology layer you leveraged > in order to build your maps/ecs. > > was that specified in the ALTO protocol? or do we have to specified what exact factor should be included? > > * the network map just includes PID and endpoint address >> >> does the information include the cell id or ap id? >> > > > it will be reflected in your pid's. > > in my understanding, the PID is a kind of self-defined naming system in ALTO, does they have the same syntax or semantics of the cell id or ap id defined in GSM or other networks? > s. > > > > > On Thu, Apr 28, 2011 at 6:40 PM, Sebastian Kiesel <[email protected]> >> wrote: >> On Thu, Apr 28, 2011 at 09:27:51AM +0800, ?????? wrote: >> > > I agree. But the creation of maps based on the (wireless) topology and >> > > other {cost,distance,performance,...}-related information as well as >> > > feeding them into the ALTO server is out of scope of the protocol >> spec. >> > > >> > why does the creation of this map is out of the scope of the protocol >> spec? >> >> Which information to put into the map depends on the network technology, >> topology, optimization goals, policies (e.g. how much information about >> the network is an ISP willing to disclose), how much effort you want to >> spend, etc. So it would be difficult to write down an one-size-fits-all >> algorithm. Furhermore, the algorithms used for creating the maps may be >> a field where different ALTO server vendors may come up with different >> solutions, i.e., market competition. >> >> For interoperability between ALTO clients and servers of different >> vendors it is important that the syntax of the map and the request/reply >> protocol is clearly specified. This is within the scope of the document. >> >> > the ALTO server should provider the information to client to make sure >> that >> > P2P application make the appropriated selection for the wireless peer. >> >> I agree. >> >> In your wireless scenario, is there any kind of guiding information >> you would like to transport from the ALTO server to the ALTO client, >> which cannot be expressed with the syntax in the current protocol >> specification? >> >> >> Thanks, >> S. >> >> >> >> -- >> Best wishes, >> >> Beijing University of Posts & Telecommunications (BUPT) >> Zhu Xiao ( 朱潇 ) >> E-mail: [email protected] >> mobile:+86 134-8881-9004 >> > > -- Best wishes, Beijing University of Posts & Telecommunications (BUPT) Zhu Xiao ( 朱潇 ) E-mail: [email protected] mobile:+86 134-8881-9004
_______________________________________________ alto mailing list [email protected] https://www.ietf.org/mailman/listinfo/alto
