[cso] Questions/Comments on Taipei_CSO_Gloabl Load Balancing Strategies_BUPT.pdf

["Mcdysan, David E" <[e-mail]>]

This looks interesting, but I had a number of questions and as mentioned
in the meeting wanted to start an Email thread on this.

Slide 12

A number of questions on the notation:

f-sub ac a is application parameter, what is c? The node selected on slide
13?

This is unitless? Can the parameter phi be interpreted as the relative
cost of memory and compute? Does memory model RAM or storage?

f-sub bc is network parameter, what is c? The path implied by the above
server selection?

Which is also unitless? Is assumption that TE weight is related to
economic cost?

Could the beta parameter be interpreted as the relative cost of
compute/storage and networking?

Slide 15

Please confirm as answered in the meeting that the y-axis of the left
graph is alpha. 

Is the traffic parameter on the x axis the relative load offered to the
topology? Is there any way in which the traffic is selected and/or the
topology and its capacity determined based upon the traffic pattern?

Is there a higher-level optimization in terms of selection of the capacity
allocated at the candidate server sites and/or the topology and its
allocated capacity? Or have you selected a topology and candidate sites
based upon some existing network or a published reference?

Thanks,

Dave