Post reply

perf question

  • September 12, 2017 at 6:34 am

    #339067

    Hi All,

    I am little confused between baselining and benchmarking performance in SQL Server. Is both terms mean the same or they can be interchangably used? Little confused. Can someone please little explanation in simple terms?

    Thanks,

    Sam

  • September 12, 2017 at 6:48 am

    #1959080

    vsamantha35 - Tuesday, September 12, 2017 6:34 AM

    Hi All,

    I am little confused between baselining and benchmarking performance in SQL Server. Is both terms mean the same or they can be interchangably used? Little confused. Can someone please little explanation in simple terms?

    Thanks,

    Sam

    This is one those cases when any two people you talk to may have differing perspectives on the meanings of those terms.   For me, a "baseline" is a measurement of something, with the objective of eventually measuring something else using the same methodology, so that the results of those measurements can be compared.   A "benchmark" is usually a performance measurement, and usually of some integrated system as opposed to any one component, with the goal of being able to compare dissimilar systems based on each system's performance benchmark.   It's usually a broader measurement when you're referring to a benchmark, but is not necessarily any less "lab-like" in terms of the detailed level of measurement involved.   I don't think the two terms are equivalent, as they refer to different, but similar things.

    Steve (aka sgmunson) 🙂 🙂 🙂
    Rent Servers for Income (picks and shovels strategy)

  • September 12, 2017 at 7:09 am

    #1959087

    vsamantha35 - Tuesday, September 12, 2017 6:34 AM

    Hi All,

    I am little confused between baselining and benchmarking performance in SQL Server. Is both terms mean the same or they can be interchangably used? Little confused. Can someone please little explanation in simple terms?

    Thanks,

    Sam

    I would look at a baseline as something you can use to spot abnormalities in your system.  To get a baseline, you might need to monitor a system for days, weeks, or longer to be able to see what the "normal" workload looks like.  Once you've got that, then you can spot abnormal behavior.
    An example might be, you record and track your servers Batch Requests/sec for a couple months and graph it out, including an average of the data.  Then you notice the counter is going well above the average for long periods, rather than the occasional blip.  Now you know you have something to investigate, whereas without a baseline you might be wondering if this is just normal behavior.

    A benchmark, generally, is a one-time measurement of "what is the max this system can handle / do, performance-wise."  At least, that's how I've always looked at it.

  • September 12, 2017 at 1:07 pm

    #1959172

    Hi Jason,

    Thanks, I got it.

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

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