Last modified 6 years ago Last modified on 02/01/2013 01:47:00 PM

Database Scalability in DC3b

Detailed plan

  • demonstrate near neighbor query on 2-3 nodes (query broken into subqueries, multiple databases, subqueries run in parallel, data partitioned by hand, no cross-server joins, results streamed to client, using xrootd, minimal tools for partitioning). [mostly DW]
  • architecture design
    • run near neighbor query on lsst10, test performance (includes porting to linux) [JB]
    • research htm/healpix/stomp/dif [SM]
    • research xrootd vs gearman
  • scalability tests, benchmarking (1st round, 25 nodes)
    • scalability related fixes
  • partitioning
    • implement generic query partitioner [SM]
    • basic, generic query parser in place ("select from where", supports queries that select single row, select from multiple partitions, join with non-partitioned tables) [DW]
    • 06/15/10 Implement data synthesizer [DW]
    • 06/15/10 Demonstrate queries involving partitioned and non-partitioned tables (via shared volume) [DW]
  • mysql proxy [JB]
    • API between the proxy and the master [DW, JB]
    • extracting and passing hints
    • fetching results
    • concurrency
    • 06/30/10 ellipse, polygon, circle, "objectId IN" [JB]
  • 06/30/10 sql functions to support spatial constraints (ellipses etc) [SM]
  • DC3b syntax parser and aggregation
    • syntax support [DW/JB]
    • collating query results [DW]
    • aggregation (sum, average, group by, sort by, ...) [DW]
    • query parser: support aggregation {DW], basic geometry support (bounding box and circle) [SM]
  • 05/15/10 task manager [KT/DW]
    • rewriting backend with new xrootd
    • receiving and running multiple queries simultaneously
  • 07/30/10: data release tools [JB/?]
    • complete tools for partitioning data [SM/JB]
    • implement admin tools for releasing data (reindexing, marking read only, etc)
    • implement monitoring of the system (for troubleshooting/debugging)
  • 06/30/10: first release
    • returning results
    • threading in c++
    • multi-queries
    • short queries
    • 06/15/10 generating objectId,chunkId table from partitioner [SM]
    • 07/15/10 preventing obvious killer queries (result size limits) [DW]
  • 06/01/10 - 07/31/10 deployment and troubleshooting on lsst10 and 2 slac servers
  • documentation
    • 06/30/10: for users [DW/JB]
    • 07/30/10: for all hands meeting [DW/JB]
    • 11/01/10: for PDR [JB/DW]
  • 08/31/10 scalability tests and benchmarking (2nd round, 100 nodes)
    • including redoing test with 4 and 25 nodes
    • cluster management tools
    • tests with and without compression
  • 08/31/10 adding support for advanced sql syntax
    • order by, limit, having etc
  • interactions with users
    • 09/30/10: basic interactions (query status)
    • 12/31/10: task manager (location of query results, query progress/cost)
    • 12/31/10 mysql admin commands (for example show processlist)
    • 12/31/10 statistics

Beyond DC3b

Tasks left out for DC4:

  • 03/31/11 task manager (killing/suspending jobs)
  • 06/30/11 shared scans
  • inter-node communication (near neighbor query with distance larger than pre-calculated adjacent regions)
  • fault tolerance
  • multi-front end (for scalability and fault tolerance)
  • inserts and deletes (needed by Association Pipeline and unreleased catalog)

After DC4

  • semantic analyzer (checking validity of table names, column names etc)
  • detecting slow chunk-queries
  • sql subqueries

Hardware needed

see db/DC3b/Hardware