Results 1 to 3 of 3

Thread: How to reduce overhead at source (SQL Server 2012)?

  1. #1
    santiagovm is offline Junior Member
    Join Date
    Jan 2015
    Posts
    2
    Rep Power
    0

    How to reduce overhead at source (SQL Server 2012)?

    Ran some tests to get a sense of the overhead that Attunity Replicate creates at the source database server and got some concerning results.

    For example, inserts at the source take 33% longer while table is being full loaded when compared to no replication running at all. That is a considerable impact in the production system.

    Also, updates take 54% longer while table is being full loaded and 43% longer while table is under change processing.

    Is there a way to reduce the overhead at the source database?

    -Santiago

  2. #2
    stevenguyen is offline Senior Member
    Join Date
    May 2014
    Posts
    301
    Rep Power
    6
    Hello Santiago,

    What is your Target DB?

    What is the configuration of your task?
    - How many tables?
    - any filter or transformation?

    Where are you getting your % from?

    What version of Replicate are you running?

    Thanks,
    Steve

  3. #3
    santiagovm is offline Junior Member
    Join Date
    Jan 2015
    Posts
    2
    Rep Power
    0
    Hi Steve,

    The target database is also SQL Server 2012.

    The task contains one table without any filters or transformations.

    I am using two servers:


    • Server #1: source sql server
    • Server #2: target sql server and Attunity Replicate


    I am using Attunity Replicate version 4.0.0.127

    The percentage is calculated as follows:

    1) populate table at source with 150K records (this amount of records makes full load last around 80 seconds)
    2) start full load
    3) run script against source database that updates 25K records one at a time. Script finishes while full load is still running (as required for this test)
    4) record how long it took for the script to finish

    Then I run the test again without replication running, no full load, only updates happening at the source. Record the script execution time again and then compare the numbers.


    • script run #1 with replication: 47.3 seconds
    • script run #2 with replication: 70.0 seconds
    • script run #3 with replication: 61.5 seconds


    avg with replication: 59.6 seconds



    • script run #1 w/o replication: 31.7 seconds
    • script run #2 w/o replication: 37.8 seconds
    • script run #3 w/o replication: 46.8 seconds


    avg w/o replication: 38.8 seconds


    updates at source took 54% longer when replication was running than when replication was not running

    (59.6 / 38.8) - 1 = 0.536

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •