create new tag
view all tags

Meeting Minutes - 2012-11-22

Attendants: Ziyan, Gang, Fabio

Time: 2:30 p.m.

Secretary: Fabio

Progress report

Gang reports on the progess made since the previous meeting. His presentation is attached.

  • Gang verified that the MySQL indexes included in the dump were actually created and used by his tests. In addition, he set to ON the option innodb_file_per_table so that each table has its own file and associated indexes. This setting improved the performance about 10% (see slide 4 for precise figures). From this work we can conclude that the discrepancies in performance observed by Gang's tests and the performance of the LHCb DIRAC portal when creating accounting plots are due to the hardware configuration of the machines.
  • Gang downloaded and installed Cassandra in a single node in his test machine. He is currently learning the concepts introduced by Cassandra and the possibilities it provides for data modelling, such as column families, super-column, etc. and how they can be exploited for modelling the information about the DIRAC accounting records.

Next steps

We agreed in the next steps:

  • Gang proposed to make a test which involves defining the minimum set of columns, super-columns and super-column families that are needed for each particular type of plot. The idea is to store in each column family only the information for that particular plot and nothing else. This schema supposes some storage overhead because some of the accounting information will be duplicated, but this denormalization may be the price to pay for getting good performance results when querying the data. We agreed to explore this approach with the 5 most frequently used type of accounting plots. Gang will report on his results with this test on next meeting.
  • For doing this test, Gang will use one BES-III machine which has a better hardware configuration that the one used so far. The tests using MySQL will be performed also on this same machine to collect baseline performance figures which will be ompared to the results obtained using Cassandra for generating the same plots.

Next meeting: December 6th, 2:30 pm, Fabio's office

-- FabioHernandez - 2012-11-26

Topic attachments
I Attachment History Action Size Date Who Comment
PowerPointpptx discussion(11.22).pptx r1 manage 411.0 K 2012-11-26 - 04:31 FabioHernandez  
Topic revision: r1 - 2012-11-26 - FabioHernandez
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback