[evla-sw-discuss] Delay models to station boards

[Bill Sahr]

Pls see below.  bill

Kevin Ryan wrote:
> Models should be addressed by Antenna/baseband not by CMIB; the MCCC  
> affords that.  We (I think all of us?) agree that Barry's worst-case  
> would overly tax MCCC.  Since the Station Board to Antenna is a  
> physical thing and since this relationship is mapped in the  
> parameters database that Barry mentions, then it should not be a  
> problem for the Executor to still send models to Antenna/basebands  
> and have them arrive at the proper CMIB.
> 
> And this can be done either by unicast or multicast UDP or via the  
> current REST/TCP interface that the CMIB uses now to receive models.
> 
> The suggestions offered in this thread to cover for UDP's dropped  
> packets involve work for the CMIB that it should not have to do, and  
> only report the problem without fixing it (assuming that if the CMIB  
> did discover and report a dropped model, it is already too late to  
> resend it).
> 
> Barry's worst case should not burden REST/TCP if the models don't  
> have to funnel through MCCC.

Barry's worst case: For 27 antennas, 4 station boards/antenna, OTF
mapping at a high frequency - 600 models per second.  I've seen your
(Kevin's) email on the testing for sending UDP packets (nice job,
interesting figures), but haven't seen anything for REST/TCP.

Bruce's email speaks of a max XML message size of 14KB.  So, 600
models/sec, @ 14KB per model (worst case)?  8,400KBytes/sec, i.e.
8.4 MBytes/sec ?  That's probably an overestimate.  I have little
feel for REST/TCP capabilities with regard to data rate.  Can you
supply some figures ?
> 
> Even though Barry originally wanted to "Reaffirm the long term  
> direction we want to take", I think we should wait to see how things  
> work.  I recommend that we continue with the existing REST/TCP  
> interface but bypass the MCCC.  It would serve to satiate those of us  
> who want the Berlin Wall between WIDAR and Executor to think of this  
> mapping database as a part of the VCI. :)
> 
> Kevin
> 
> On May 15, 2008, at 8:14 AM, Barry Clark wrote:
> 
>> I've made provision for storing this in the parameters database.   
>> It was
>> called for in Sonja's schema of a couple of years ago, so I put it in.
>> MCCC will need to know a lot of this sort of thing too, and I think
>> that's OK, so long as we get them from the same database.  When we
>> get to talking about code/repository sharing, I'll be urging them to
>> use our Parameters class.  I'm quite happy with it at the moment.
>>
>>
>>> From: Bryan Butler <bbutler at nrao.edu>
>>>
>>>
>>> how?
>>>
>>>
>>> Sonja Vrcic wrote:
>>>> EVLA M&C System (Observation Preparation and/or Executor) knows  
>>>> which
>>>> antenna is connected to which Station Board.
>> _______________________________________________
>> evla-sw-discuss mailing list
>> evla-sw-discuss at listmgr.cv.nrao.edu
>> http://listmgr.cv.nrao.edu/mailman/listinfo/evla-sw-discuss
> 
> _______________________________________________
> evla-sw-discuss mailing list
> evla-sw-discuss at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evla-sw-discuss