Block sizes and Oracle

I'm posting this for a friend, respond to him or post:

Wondering what experiences people have with changing the FS block sizes to increase throughput for moderately large Databases. We suspect that this change will have an effect but are unable to test because ultrix only allows 8KB blocks.

Attempting to call newfs with a "-b 16384" produces a

read error: 96
rdfs: Bad address

The stats of our env are Ultrix 4.2 on Decstation 5000/200 running oracle 6.0.33 we have data files on their own disks. Each disk is a FS and is a 665M or 1.2G. These files on the average take up 50M some are as small as 5M and some are as large as 75M.

Is it reasonable to assume we would get a increase, provided the bottleneck is in geting data off the disk?

I know what the books say I can do but we interested in that people have actually done or tried.

No, going to Raw partitions is not an option.

Regards,
Lee Parsons Brown & Root Inc. Houton, Tx

Please mail Commentary to hou13us!parsons_at_uunet.uu.net

-- 
Steve McDowell             . . . . o o o o o 		Opinions are 
Exlog, Inc.	                  _____      o     	mine, not my 
mcdowell_at_exlog.com      _____====  ]OO|_n_n__][.    	employers..
		       [_________]_|__|________)<