[alto] Comments on draft-deng-alto-p2p-ext

[Gao Kai <[e-mail]>]

Hi all,

I found all these new endpoint properties very useful and I can't help
notice many are specialized for mobile and data center scenarios. 
However I still have doubts on some issues and here are some comments:

1. I understand that the "precision" field is introduced to indicate the
typeof the content.  However, in 5 cases out of 8 the field is left
empty, in other 2 the only difference is between an exact value and the
ranking.  In the last case, "geolocation", the "precision" field can be
chosen from four different values, but according to my understanding, an
endpoint can have a geolocation for each one of them.  What if the ALTO
server wants to provide all kinds of them?  Since they all share the
property name "geolocation", it can lead to conflicts when the data are
encapsulated in a JSON object.

So instead of using "precision", I think it is better to provide 4
different geolocation properties.  The same idea can apply to
"network_access" and "provisioned_bandwidth": use "-rank" suffix in the
name to indicate the value is a ranking.

At the same time, introducing
"geolocation-type"/"network-access-type"/"provisioned-bandwidth-type" to
indicate what "precision"s are supported looks like a good idea to me.

2. The names are using underscores instead of hyphens.  However I think
it is better to keep it compatible with RFC 7285 in which property names
use hyphens to concatenate words.

3. Why not use strings to represent the bandwidth?  Such as "1 Gbps". 
It's more compact.  I'd also like to know if there are some other
considerations why the metric and value are separated.

==================

Regards,
Kai