Winnt vs. Windows 2000 performance

[smcdmc]

I am having a performance issue that I can't figure out. I am running Oracle 8.1.6 on two separate machines. Machine "A" has Windows NT with dual processors. Machine "B" has Windows 2000, 933 with 1 gig ram. We are trying to benchmark each machines performance relative to our database. Both machines are stand alone, both have one user (the testing user), both have the exact same database configuration (we copied the script from one machine to the next). On the Winnt machine we are averaging 1123 table inserts per second. On the Windows 2000 machine, we are averaging 323 table inserts per second. I am not sure what to look for if anything that can explain this discrepency. Any thoughts?

Thanks,

smcdmc

[sambavan]

here your processors seems to play a vital role. The load your NT dual can match more when compared to the load 2000 could reach. Since Windows is of Thread archetecture, the thread handling becomes better on a dual processor when compared to the single processor one.

Sam

[drain surgeon]

Don't be surprised @ the result. I don't believe you did say what speed the processors in the NT server were. But think of it this way. You have the Win2k Server with one big channel this channel can carry x amount of data.

The NT Server has two smaller channels, but added together they are bigger that the single channel. Therefore the NT Server will do more work. This could be an explaination as to the difference in performance. Althought you probably would expect better from the Win2K server.

You said the databases were identical. Does this mean the parameters set in the init.ora file are the same. Check v$sysstat to see the cpu usage. If the cpu has a high idle time try increasing the DB_FILE_MULTIBLOCK_READ_COUNT parameter.

Are you experiencing very high swapping or paging.

Could be a host of other things. If all these fail, try collecting some Hit Ratio statistics. Good luck.

The Drain Surgeon