SQL Server Stupidity



  • We have a part time network contractor who seems to think that because he is in charge of hardware he sould make all decisions for SQL Server when it comes to hardware.  His stupid point number 1.  That a 7200rpm SATA drive would give the same performance as a 15k rpm scsi drive.  2. That it would be better for SQL Server to go with less, slower, larger drives.  and last but not least 3.  It is not best practice to multi home any server. 

    I almost laughed in his face when he made these points, but at this company he got his way.  The person in charge of everything bought into his bullshit because he thinks everything is a hardware/software issue.   And this person wonders why SQL Server is running so slow.

    This person may be one of the dumbest people I have ever met.

    Just a day in my life in dealing with Dumb and Dumber as I have nicknamed them.

     



  •  Just for reference, the stories tend to sound less bitchy when you let the readers judge the WTFery.  :)   "OMG, these people are so freaking stupid" tends to come back to bite you in the ass around here.



  • I will keep that in mind.  Thanks for the advice.

     



  • Although it's obvious that 1 and 2 are not correct, I wouldn't be so quick to count them as the primary reason the server is slow.  My experience is that fixing the disk problem will probably get you an order of magnitude performance improvement at best.  Query design and indexing can easily account for a three order of magnitude performance variation.

    Depending on a variety of factors, the best way forward may well be to spend $5,000.00 on new drives.  However, it's possible that spending that same $5,000.00 on someone who knows how to performance tune a database and the application would fix the problem more effectively.  It's possible that the part-time network contractor was aware of all of these things and shifted some of the hardware budget to the development team.  It may even be possible that he miraculously saved an application that was going to blatantly fail on day one and got it to the point of a limping success.

    Also, if you believe that the person in charge who believes that everything is a hardware/software issue is wrong, then where is the problem?  Is it firmware?

     You are probably right.  But, your story isn't as convincing as you think it is.



  • Totally agree, and the performance has been measured.  We even had an outside vendor come in and measure as well.  The conclusion was we do not have enough disks handling the IOPS we put against the database.  What I meant by the hardware/softwre thing was that person thinks that the networks guys should take care of any hardware problems, even when they dont know anything about SQL Server. 

     It is just kiindof humerous now cause at most places i have been once you show enough proof on where the problem lies you would expect to at least move in the right direction to fix the issue.



  • @cHao said:

     "OMG, these people are so freaking stupid" tends to come back to bite you in the ass around here.

     

    let truth = truth + my_personal_exp

     

     New drives are awesome, and worth it! Now I can store 2TB where I used to put 2GB (in good ol' leafblower caviars!), and at orders of magnitude faster. I did like freakin awesome names Quantum gave their drives though (2TB Fireball anyone :).


  • Garbage Person

    @Jaime said:

    My experience is that fixing the disk problem will probably get you an order of magnitude performance improvement at best.  Query design and indexing can easily account for a three order of magnitude performance variation.
    This. I once fixed a query with a 66-year runtime so that it runs in under 1 second. I don't know how many orders of magnitude that is, but it's a LOT of them.

     

    Also, I have physical evidence that certain modern 7200RPM SATA (perpendicular recording, 320gb/platter) drives are nearly equals to older 15000RPM SCSI drives in addition to being vastly larger - all thanks to density. So your "stupid hardware guy" may have been right.


  • Discourse touched me in a no-no place

    @Weng said:

    I once fixed a query with a 66-year runtime so that it runs in under 1 second. I don't know how many orders of magnitude that is, but it's a LOT of them.
    It's about 9.



  • @PJH said:

    It's about 9.
     

    I almost misread "orders of magnitude" as "factor", probably so I could have made the "over nine thousand!" joke trope, but alas, alas.


Log in to reply