Re: Poor efficiency with oracle server

From: Richard D Holowczak <holowcza_at_andromeda.rutgers.edu>
Date: 8 Sep 1994 20:09:12 -0400
Message-ID: <34o938$f2u_at_andromeda.rutgers.edu>


abe_at_liber.win.net (Liber) writes:

>We are using oracle server on a dos based novell lan. We have about
>20 workstations on the lan. We are experiencing very slow
>processing with our oracle database. Someone suggested adding a
>dedicated novell server that would run oracle. We tried this but
>it didn't seem to help. Should we be looking at maybe increasing
>the power of the workstations?
 

>Any suggestions??
 

>Karen Ward

 Hi there. You're not giving us much info to go on.  Perhaps if you could supply the following info:

    File / Database Servers:

      Processor          (i.e. 80486 or Pentium)
      RAM                (16, 24 or 32 Megs ?)
      Hard Disk          (1, 2 4 or 10 Gigabyte ?)
      Network Cards       (Ethernet, Token Ring, 16 or 32 bit ?)
      System Bus         (EISA, ISA or Microchannel)
      Netware version    (3.1x or 4.x) 

      Any other applications running on the "Database" server ?

   Client PCs:
      Processor (i.e. 80486 or Pentium)
      RAM
      Network Cards
      Operating System (MS-DOS, OS/2, Windows, Windows for Workgroups)

     Type of applications  i.e. SQL*Forms for DOS or Forms for Windows
   

     In general, are there any other apps using the 
     network that are causing a great deal of network traffic ?

For a start, Oracle on Netware likes a lot of RAM. I've got 32 Megs of RAM on a 486 machine for 30 users. If you have some other applications eating up network bandwidth, response time may be slow. Underpowered client machines (i.e. running Oracle Forms 4.0 with 8 Megs of RAM) can make the server appear slow.

I've also heard Oracle 7 under Netware 4 is about 10 - 25 % slower than Oracle 7 on Netware 3.12 although I have not seen this for myself.

To me, these are the obvious things to check on.

Next I would get a good book on tuning Oracle databases and have a look at the data model you have, but that's a bit deep for the present . . .

Hope this helps

Rich Holowczak
Rutgers University
holowcza_at_andromeda.rutgers.edu Received on Fri Sep 09 1994 - 02:09:12 CEST

Original text of this message