| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.server -> Re: Oracle ODBC performance SUCKS!@
In article <354a4cf3.0_at_news.webt.com>, bubba <brittonb_at_webt.com> wrote:
>Two identical tables: one in MS SQL Server, the Other in Oracle 8 NT.
>(NT Server 4.0 SP3, 96 MB RAM, PII 266, SCSI Wide 4-g. Stripped system of no
>other extraneous applications, processes.)
>
>Access 97 linked tables via ODBC (BOth Microsoft ODBC drivers, one for
>oracle, the other SQL Server.)
>
>Loading a page full of data from the Oracle 8 server causes my one-user load
>NT server performance to sky rocket to 97% for three seconds just to pull
>back a pagefull of data in MS Access via ODBC from the Oracle 8 Server.
>
>Using the same linked table(exact schema) but MSAccess to MS Sql Server
>results in a page of data coming in less than 1 second and NT performance
>monitor shows 2% utilization on the processor for this test.
>
>Now I know that this is informal and not scientific. But , Oracle Sucks!
>
>Every time I hit page down in MS access via ODBC to Oracle 8 it takes three
>to four seconds per page of data. In MS Sql server it's pretty much
>instantaneous.
>
>By the way, I ran each test on the server while only one database server
>process was running(ie., Oracle 8 Alone on the server, then MS SQL Server
>alone on the server.
>
>What the hell is going on?
>
>How can somebody reliably develop with ODBC and Oracle 8? With 1 user Oracle
>8 is brought to it's knees!
>
>I have Oracle 8 running in the default configuration but with 30mb SGA.
>
>I absolutely have to know if I can tweak Oracle 8 into a usable system for a
>VB5 to Oracle 8 application via ODBC. If not, I have to explain to my
>customer to switch to MS SQL(they already own oracle 8).
>
>I can't imagine running a 100-user + system on Oracle 8. What's all the
>hubbub about Oracle 8 anyway. Does it just suck on NT or what?
No, it just appears to suck when you extrapolate a single user test to a 100 user system.
Oracle has a much more sophisticated buffering system, and a superior locking strategy, so the more users you have and the more potential conflicts their data requests have, the bigger the win with Oracle. It is more expensive, and does required some sophistication with tuning, even for a simple system.
Picture the difference between a high-performance motorcycle and a bus.
Of course the motorcycle can go faster. But which would you want to use to
transport 100 people and all their luggage from New York to Chicago?
>
>brittonb_at_webt.com
>
>
--
These opinions are my own and not necessarily those of Information Quest
jgarry@eiq.com http://www.informationquest.comhttp://ourworld.compuserve.com/homepages/joel_garry "See your DBA?" I AM the @#%*& DBA! Received on Tue May 05 1998 - 14:18:29 CDT
 |
 |