As Mike correctly points out, it seems that the choice of who owns
the 4 day disk
buffer at the mountaintop is one of the main drivers to decide where
to divide
responsibility between camera DAQ and DM. Our common goals are to
1. Insure science data is not lost
2. Build an integrated, supportable system at minimal hardware cost
3. Minimize the effort (cost) to write/debug all the software
4. Properly integrate the DAQ and DM with the other required control
systems on the mountain
5. Minimize hardware/software support costs over the life of the
telescope.
We shall continue our discussions at SLAC later this week.
John Thayer and I were the only ones on the tech assess concall this AM.
Next concall is Monday, March 20, 10AM Pacific time. We will
continue to discuss the
DAQ/DM architecture and technologies to move data down the mountain.
Don Dossa
On Mar 3, 2006, at 5:09 PM, Huffer, Michael E. wrote:
Terry:
Stuart, Don, and I had what I believe was a very productive meeting
today. This is my (very) short synopsis of that meeting. Stuart and
Don will, I'm sure, chime in and either amplify or correct any of
my statements below:
- We first spent some time communicating our two very different
visions of the networking infrastructure necessary to support
camera DAQ. It was very helpful (at least to me!) to understand
Don's vision and the motivations which drove him to make his design
choices.
- We then focused on (in terms of the networking infrastructure)
attempting to define the boundaries between the CCS DAQ system and DM.
- I think its fair to say Don would much prefer that this boundary
by drawn much tighter then as outlined in my proposal and certainly
CCS should not penetrate not outside the mountain. In terms of my
proposal, his two suggestions would be: 1) between RNA and switch
or 2) between switch and the down-the mountain facility. He would
also maintain that the 4-day buffer belongs to DM.
- We discussed the merits/demerits of each approach without coming
to any real conclusion except that we need much more time to
discuss these very issues! However, clearly, the decision on who
owns the 4-day buffer is a strong driver on the mountain-top
network infrastructure.
- Stuart then explained the importance of the decision on what to
do about the cross-talk correction. Don understands the increased I/
O burden this places on the base-camp and would like to think some
more about the consequences before he comes to any conclusion.
- Last, we spent some time discussing the idea of a homogenous
networking solution and should that common solution be IBA? I feel
as if we didn't come to any concrete conclusion, but maybe Stuart
and Don will correct me.
Don will be at SLAC for the director's meeting and we will meet
further on Thursday (presumably with Tim and Jeff). --- mike
_______________________________________________
LSST-data mailing list
[email protected]
http://www.lsstmail.org/mailman/listinfo/lsst-data
