Computing Division Requests The following is a list of tasks about to be officially submitted to the Computing Division for requested support. D0 personnel must remain responsible for defining task requirements, evaluating application design, and monitoring application deployment. Are we comfortable ‘farming out’ these tasks? Are there other tasks to be considered?
Comp Div requests, cont’d Data Logger, Part 1 Design and develop a system to: –interact with a Run Control application –interact with COOR –receive events over network from L3 –log events in partition files Patterned on DART logger Data Logger, Part 2 As a component of the RIP project, design and develop a system to transfer the contents of partition files to either: –remote FCC RIP facility –local tape facility Provide RIP flow monitoring tools
Comp Div requests, cont’d Run Control Design and develop a system to: –provide central run control management (server) –provide GUI (client) –interact with COOR, Data Logger Event Distribution produce the framework within which selected events (by run, trigger) are received from Level 3 and distributed to client nodes provide both server (with similar functionality to Data Logger) and client ends –Digital UNIX server –UNIX, Linux, NT(?) clients provide system monitoring functionality
Comp Div requests, cont’d Database help specifically the problem of replication of locally created and entered data to the remote reconstruction / analysis facility generally as consultants for –db administration –db performance optimization GUI development support Python GUI support multiple platform support user interface design consultation Online system administration expected that Online system manager will be a Computing Division person assigned to work within D0 responsible for Digital UNIX, Linux, and NT components
Comp Div requests, cont’d Control and monitoring jobs provide user interface to High Voltage program –support existing HV front end system provide user interface to other EPICS- based monitoring applications
Beams Division Requests We are also planning to make an official request to the Beams Division for assistance in several areas. This isn’t as far along, so comments on possible tasks are welcome. We should also be doing some ‘back door’ negotiations to determine extent of interest on collaborative tasks.
Beams Div requests, cont’d Accelerator Console Run 2 version of the ACNET console Integrated with D0 Online system? Information exchange Tools to get information from the accelerator system –luminosity monitoring –beam position monitoring Tools to send information to the accelerator system –feedback of beam position –feedback of beam quality Control system tasks 1553 driver for EPICS continued support of Shea/Goodwin systems others???
InterProcess Communication We’ve been working on the task of producing a communications package for over a year… first efforts were to survey existing packages; in general we thought such were overly complex for our needs; we had d0ip (from Jan Hoftun) as a starting point for a home-grown solution; but wanted a number of features beyond a simple socket-level package: C++ class structure, flexible (TCP, UDP, multicast), reliable, efficient (with multi-threaded implications), and easy to use… The more we’ve worked, the more we’ve seen that any home-grown solution would be as complex as externally available packages and probably more error prone, less efficient, and less full-featured
IPC cont’d Our suggestion is thus to revert back to the idea of adopting a publicly available package, ACE the ‘baby’ (well grown now) of a professor at Washington Univ, St Louis seems to be supported by a number of his students/staff (commercial support also available) in use at a large number of institutions Laura will tell you about the local installation of ACE. I’ll finish with some comments on what to do next.
ACE issues Needs: Quickly do all of the below… Understand, in detail, the operation of the following components: –IPC SAP package –threads wrapper –higher level patterns: reactor, etc Determine interaction with existing D0 efforts (Scott, Gordon) Determine which ‘value added’ D0 features are still required –buffer management? –Message retransmission? Document D0 usage of ACE –publish the API