Data360 Analyze

 View Only
  • 1.  Server Performance Report

    Posted 10-10-2023 05:44

    I am wondering if anyone in the community has ever built a dataflow to track what impact different runs may be having on the server?

    I would like to be able to see the actual compute of each node within each dataflow for each run, both adhoc and schedule. If actual compute is not possible is there any sort of proxy that could be tracked and measured to help us identify poor performing runs on our server?

    Any help on this would be great.


    Toby Harkin
    Telstra Corporation Limited
    Sydney NSW

  • 2.  RE: Server Performance Report

    Posted 10-10-2023 06:53

    Sidetracking a bit but,

    The lookup node can have a big impact on the ETL and server performance, it can bottleneck.

    "The Lookup node is recommended for use with a small data set on the right input as the entire right data set is loaded into memory", so in those cases its better with a join/merge, or increasing the java heap space, if it doesn't compromise the server's available  memory.

    Also, if doing heavy computing on a large datasets, the performance gains from calculate fields instead of a transform node can have a huge impact on the node processing time.


    Henrik B
    E.ON Sverige

  • 3.  RE: Server Performance Report

    Posted 10-12-2023 06:45
    Edited by Adrian Williams 10-12-2023 08:02

    The base information for the performance statistics you are looking to track will be included in the system's  <site-7731>/logs/lae-audit.log  file and, depending on the timeframe for the analysis, any 'rotated' audit logs (which will be named  lae-audit.log.yyyy-mm-dd).

    While the following use case does not use a data flow, you may want to consider using the 'jq' command line tool to analyze the lae-audit.log file. See the following articles relating to its use.

    How to use jq on linux Environment

    Data360 Analyze: How to use jq on Windows

    Alternatively, you could build a custom data flow which parsed the JSON audit log event information.

    Adrian Williams
    Precisely Software Inc.