Thread

  1. postgres performance

    Shiby Thomas <sthomas@cise.ufl.edu> — 1998-01-15T18:01:37Z

    Hi,
    
    I have installed PostgreSQL6.2.1 from my own user account (I was just trying
    to test out something first). It seems to be running too slow. For example,
    the self join of a table with roughly 300K records takes 2-3 hours. There is
    an index on the join attribute.
    I am running it on a 8 processor (each is a 248 MHz SUNW,UltraSPARC-II)
     machine with a total of 2.0 GB main memory.
    But, since it is not parallelized, it may get the power of only one processor.
    
    Does it make a difference in performance since I have installed it from my
    user account, and not the root. Also, I compiled it using gcc on a 
    solaris 2.5 machine and the machine I ran it is a solaris 2.6 machine. I did
    that since the solaris 2.6 machine is the fastest we have here and the 
    executable
    compiled on 2.5 was running properly also. Can that have an impact on the
    performance?
    
    Thanks
    --shiby
    
    
    
  2. Re: [HACKERS] postgres performance

    Marc G. Fournier <scrappy@hub.org> — 1998-01-15T18:15:59Z

    On Thu, 15 Jan 1998, Shiby Thomas wrote:
    
    > Hi,
    > 
    > I have installed PostgreSQL6.2.1 from my own user account (I was just trying
    > to test out something first). It seems to be running too slow. For example,
    > the self join of a table with roughly 300K records takes 2-3 hours. There is
    > an index on the join attribute.
    > I am running it on a 8 processor (each is a 248 MHz SUNW,UltraSPARC-II)
    >  machine with a total of 2.0 GB main memory.
    > But, since it is not parallelized, it may get the power of only one processor.
    > 
    > Does it make a difference in performance since I have installed it from my
    > user account, and not the root. Also, I compiled it using gcc on a 
    > solaris 2.5 machine and the machine I ran it is a solaris 2.6 machine. I did
    > that since the solaris 2.6 machine is the fastest we have here and the 
    > executable
    > compiled on 2.5 was running properly also. Can that have an impact on the
    > performance?
    
    	There may be optimizations in the 2.6 libraries that would improve
    performance, but I wouldn't suspect that it would make *that* big of a
    difference.  What is your SQL/join statemnt?  How are you running
    postmaster?  What does 'explain' show?
    
    
    
    
  3. Re: [HACKERS] postgres performance

    Bruce Momjian <maillist@candle.pha.pa.us> — 1998-01-15T18:23:07Z

    > 
    > Hi,
    > 
    > I have installed PostgreSQL6.2.1 from my own user account (I was just trying
    > to test out something first). It seems to be running too slow. For example,
    > the self join of a table with roughly 300K records takes 2-3 hours. There is
    > an index on the join attribute.
    > I am running it on a 8 processor (each is a 248 MHz SUNW,UltraSPARC-II)
    >  machine with a total of 2.0 GB main memory.
    > But, since it is not parallelized, it may get the power of only one processor.
    
    You said a self-join.  I think we have a performance problem there. 
    Vadim?
    
    
    > 
    > Does it make a difference in performance since I have installed it from my
    > user account, and not the root. Also, I compiled it using gcc on a 
    > solaris 2.5 machine and the machine I ran it is a solaris 2.6 machine. I did
    > that since the solaris 2.6 machine is the fastest we have here and the 
    > executable
    > compiled on 2.5 was running properly also. Can that have an impact on the
    > performance?
    > 
    > Thanks
    > --shiby
    > 
    > 
    > 
    
    
    -- 
    Bruce Momjian
    maillist@candle.pha.pa.us