Re: [nm-wg] standard timestamp format

[Dan Gunter <dkgunter@lbl.gov>]

> > 2. Single-source timestamps
> >   (a) 64-bit counter (?)
> >
> > Then you are suggesting that a #2 timestamp could be optionally  
> > included wherever we allow a #1 timestamp.
>
> Yes, and the units should be nanoseconds in a 64bit counter with  
> arbitrary zero point. The value is
> quantized to whatever resolution the provider supports, e.g.  
> milliseconds for java, 10ms ticks for
> Linux, about 1us for Solaris.
I would suggest that the units be an optional parameter, since they 
aren't always needed, and maybe aren't always known, i.e. in the case of 
some kind of hardware clock.

> There also needs to be a definition of the parameters that combined  
> together make up the signature of
> the source. e.g. the pair of IP:port endpoints identifies a unique flow  
> for TCP/IP, whatever makes sense
> for the network measurements in this case should be clear.
Good point. The source is what we are calling a "node", so I would say 
that we should include the single-source timestamp node(s) in the 
metadata along with the path endpoints and tool versions, etc. This same 
metadata is where I would put the (optional) units.

Sound good to everyone else? I can pretty easily modify the current 
report schema to do the above things. Maybe we should talk about it on 
the next conference call (hint, hint).

-Dan