Just what do you think you're doing, Dave?
Mon 17 May 2010
What is 45 days old, has visited 22 countries and already has close to 500 children but will outlive them all? Dave the ATLAS dataset. Born on the 30th March 2010 to his proud parents, the ATLAS detector and the LHC, Dave weighed in at just over 3 Terabytes. Very soon after that he was sent to RAL and the team there have been tracking his progress through the grid world ever since.
Dave is the result of accelerating two beams of protons in the LHC and colliding them in the centre of the ATLAS detector. He is a selection of the collisions chosen by the triggers at ATLAS to be of interest to physicists. As soon as Dave was packaged up he was winging his way to RAL via the dedicated 10gigabit/s links between the UK and CERN. Once he arrived at RAL he was copied into the CASTOR Storage system and registered in the ATLAS LFC, ready to be processed by scientists from across the world.
It didn't take long and within 8 days Dave already had 51 descendants. These are the datasets resulting from scientists using Dave to examine their physics models. These are then available for further analysis. Trying to keep a track of all of Dave's children (and grandchildren) isn't easy but thankfully they currently break down into 3 types; Ursula, Dirk and Valerie.
- Ursula: These are datasets that have been created by individual ATLAS users
- Dirk: Datasets that have been produced as part of the standard ATLAS central production system
- Valerie: Produced and used by the central ATLAS production system to validate a site's ability to analyse various configurations of data analysis
Brian Davies at RAL is the man behind the idea to track a single dataset and he is happy with how it has gone so far, "One of our tasks is to follow data distribution and to see how it if follows the VO's computer model. Dave was the first custodial raw dataset that RAL was responsible for from the 2010 7TeV run of the LHC. Tracking Dave is not just for fun: one of the aims is to answer the question "From one run at the LHC, how many files and what volume of data will get produced?". The thing which has been the most enlightening has been how quickly the number of real ATLAS users who have analysed Dave and his children has grown. Besides that I am interested to see how quickly his children become obsolete (if at all) and once this happens how quickly they are removed from the grid."
You can follow Dave's progress along with that of his kids at the GridPP storage blog here:
http://gridpp-storage.blogspot.com/search/label/Dave
© Copyright GridPP
If you wish to reproduce this piece please credit GridPP and contact Neasan O'Neill to say you are using it