perfsonar development work

["Jeff W. Boote" <>]

Martin Swany wrote:
> Hi Luis, All,
>
> I think this supports my argument.  The best we can hope for is to be
> descriptive (as opposed to prescriptive.)  So, FCCN can express that
> in a "Units" parameter.  We still need to decide on "b/s" vs "bit/s", 
> "bits/s",
> "bps" etc.  Then an MA can decide whether to implement a "bit/s" to "B/s"
> transformation (or vice-versa) or not.

I agree that from a protocol perspective, the most important thing is to be 
non-ambiguous.

From an interoperability of software and maintenance perspective - I think it 
is critical that we minimize the number of allowable units we use as much as 
possible. So that clients (and/or transformation services) can reasonably 
support the full set.

So, I think we are really in agreement here. We just need to articulate it in 
normative text along with the schema/name-space profile for 'utilization'.

I think having 2 types, one for bits an done for bytes is reasonable. (Don't 
really care about the "b/s" "bits/s" debate, just pick one.)

It *might* even be reasonable to add SI multipliers. (Hmm... base 10 or base 2 - 
did I say it would be reasonable?)

jeff