Compatibility level performance

  • I just restored a 2012 backup on a 2017 server and ran a rather complex statement, using different compatibility levels. Here are the times it took to execute the statement:

    Compatibility 2017: 36s
    Compatibility 2016: 36s
    Compatibility 2014: 26s !!!
    Compatibility 2012: 44s

    Is there any sensed reason why there would be a significant improvement in performance when using 2014 compatibility level?
    How can I achieve the same performance using 2017 compatibility level?

  • cmartel 20772 - Friday, March 15, 2019 12:53 PM

    I just restored a 2012 backup on a 2017 server and ran a rather complex statement, using different compatibility levels. Here are the times it took to execute the statement:

    Compatibility 2017: 36s
    Compatibility 2016: 36s
    Compatibility 2014: 26s !!!
    Compatibility 2012: 44s

    Is there any sensed reason why there would be a significant improvement in performance when using 2014 compatibility level?
    How can I achieve the same performance using 2017 compatibility level?

    It depends on what the query does and what the difference are in the execution plans. Did you compare the plans?

    Sue

  • As far as I can tell, the 2017 plan is identical to the 2014 plan (and isn't expected?)

  • I would double check the plans are the same first. Not they are not expected to be the same. They vary by server config like trace flags and by version. I would then attempt to tune it or get MS on the line to explain the results.

Viewing 4 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic. Login to reply