EISCAT & DIRAC Technical Discussion Meeting (18 Dec 2015)
Andrei gave a presentation about DIRAC, focus on DIRAC data management services. DIRAC is one of three technologies supporting LHC, has been responding for LHCb data handling. DIRAC services are based on their long-time real experience in handling large data, providing mutual solutions for managing workload, heterogeneous resource environment, HPC interaction, replication, synchronisation, bulk data transferring, etc.
Questions and discussions are followed:
Ake: how DIRAC address access to storage
Andrei: DIRAC doesn't have control on resources, it would be good to have access in-align with storage access control, however, in practices it is difficult. DIRACT addresses the problem case by case.
Yin: We have considered to adopt dCache. What is the relationship between DIRAC and dCache?
Andrei: dCache is one of storage technology which DIRAC can connect with.
Yin: We will need visualisation services, how it can be integrated into DIRACT framework?
Andrei: A lots of different visualisation services can be plugged in e.g., as part of workflow
Yin: EISCAT 3D would want to support users to process their data by submitting own parameters, can DIRAC support this?
Andrei: DIRAC at the beginning was mainly design for workflow management, which has been support various LHCb users experiments. Special technology is provided to improve performance for processing large data.
Yin: how much efforts would be needed for installation and customisation of DIRAC?
Andrei: If only for try the technology, there are ready to use setting.
Yin to all: what is our next step?
Ake: we need to have data model to be completed
Andrei: DIRAC has been supporting other communities such as CTA in data model design from the beginning. DIRAC would be happy to help EISCAT to review the data model and evaluate if it is feasible for implementation.
Ingemar: EISCAT is working on data model, will have a second workshop around the week of 11th Jan. Then we can have a meeting with Andrei.
Yin very concerns about the progress of Portal development. Since we need to deliver in Feb 2016, we urgently need solutions. Yin suggests to start to try DIRAC in Jan. Adopting existing advanced technology would avoid reinventing wheels, also save a large amount of efforts/costs in developments. By using DIRAC ready to use file catalogues and metadata services, we may have a prototype for the deliverable in Feb 2016.
There are minutes attached to this event.
Show them.
