[alto] Review of draft-randriamasy-alto-cost-context-01

[Yichen Qian <[e-mail]>]

Dear authors and all,

I think it's interesting to introduce wireless scenario into ALTO. Here's my review of this draft.

The draft introduces the cost calendar in providing the contextual cost (cell load), while it doesn’t elaborate how to use the calendar. One possible way is to use the calendar as a reference to do the prediction for the decision, while I wonder if it is accurate and useful for the user as much movement is involved in the wireless scenario and the cell load can be unpredictable.

Some detailed comments: 
1.  Introduction
“This draft brings a use case where providing different values for a same cost metric can help in optimizing the application path selection.”
- a same -> the same

“The present draft proposes to extend the cost information specified in [RFC7285] by providing, for a same cost metric, several possible cost values.”
- a same -> the same

“Previous ALTO WG discussions have suggested to introduce "the ability to "name" cost maps …”
- suggested to introduce -> suggested introducing

2.1.  Use Case 1
“The UE thus can advatageously combine the non-real time ALTO information with the real-time UDI provided by the LTE network.”
- advatageously -> advantageously

3.  Required ALTO extensions
“The aforementionned use cases can be supported with a few simple extensions to the ALTO protocol.”
- aforementionned  -> aforementioned

“these features would extend current proposals in the WG,that could be added to [draft-ietf-alto-performance-metrics]”
- missing a space before “that”

4.  Design options and examples
“Context attributes taking numerous or unpredictible values should be handled as values properties or metrics expressed in constraints.”
- unpredictible -> unpredictable

4.2.  Example IRD
“The values of 'RFcost' are provided as a an ALTO Calendar as specified in [draft-ietf-alto-cost-calendar-00] in shorter time intervals.“
- as a an ALTO Calendar -> as an ALTO Calendar

“In addition they differ, depending on the	the context values "uda" and "udna”.”
- on the the context -> on the context

“Besides, the IRD provides metric 'routingcost' as a MUST specfied in [RFC7285]”
- specfied -> specified

4.3.  Use case 1
“As a Network Map may cover a large number of cells, the Filtered Cost Map Service can be used to reduce data exchange and get information on a restricted number of cells, say PID1 and PID2.”						
- I think Cell1 and Cell2 may be more accurate here since they have been named “Cell-k”

“In this scenario, C1 is limited by its uplink capacity, C2 is limited by its downlink capacity.”
- I think the naming here for the cells is confused as there are both “Cell1” and “C1” in this section. I suppose they refer to the same thing and can be represented in a consistent way.

4.4.2.  Network Map
- The request above has four (“src”,”dat”) pairs (Cell 1->1, 1->2, 2->1, 2->2), while in the response there are only cost maps of two pairs (Cell1->1, Cell2->2).

4.5.1.  Use case 2
“The ALTO Server will provide costs w.r.t. the AND combination accross and within arrays.”
- accross -> across

5.  Deployment case
“To maintain scalability, the ALTO coverage network zone can be decomposed in one "local"ALTO Server part covering a restricted local network zone”
- missing a space before “ALTO”

Best wishes,
Eason