Attendees:
Stephanie, Debbie, Kuhkee, Sheng, Terri, Ernest, Till, Antonio, Amedeo, Chris, Diane
Overview of feedback message traffic and counts on FCOM (multicast network):
- Till suggests grouping all detectors per IOC into one packet/group, sending actuator commands as individual packets/groups.
- Controller IOCs can be placed anywhere, location will have little impact on latency. Possibly we create a 'bank' of controller IOCs attached directly to the core ...
- Antonio suggests we consider the impact of using unicast for our network, rather than multicast
BLD discussion:
Discussed the data needed by the Photon physicists, data available from the FCOM network, possible methods of getting the data from e-beam to photon networks. Some definitive statements:
- E-beam group wishes to minimize the burst packet count on FCOM multicast network
- Photon group wishes to receive e-beam data in their format
- Photon group wants the e-beam timestamp (incl.s pulse id) - this is available with every packet on the FCOM network.
Good start to what will be a collaborative development effort - need to set up working-group meetings.
FCOM ICD:
- We'll need a map file of device name<->FCOM ID. Every IOC involved in Feedback will need to load this file.
- Only need to update this file when a new device (detector or actuator) is added to the network.
- Feedback loop controllers must 'reserve' some IDs for new feedback loops, to assign to states. 'Reserved' databases are already planned for new transverse feedbacks, the IDs will be reserved for these too.
- a 'version' id will be part of the FCOM protocol to allow FCOM to check that all IOCs are communicating with the same version of FCOM.