Jared has kindly pointed out to be that the first paragraph is less than clear. I hope that I can clear it up without confusing myself further.
First, a few terms to define. This is from *Oracle(r) Database Performance Tuning Guide, chapter 8 I/O Configuration and Design.*
"Stripe depth is the size of the stripe, sometimes called stripe unit."
"Stripe width is the product of the stripe depth and the number of drives in the striped set."
"On low-concurrency ... systems, ensure that no single I/O visits the same disk twice." *My interpretation of this is that if you have few users then you want a small stripe depth so that the I/O requests will be satisified with as many disks as possible participating only once. The stripe width is not really the issue. It may be large because of the number of drives participating in the stripe set.* ** "In a system with a high degree of concurrent small I/O requests, such as in a traditional OLTP environment, it is beneficial to keep the stripe depth large." *My interpretation of this is that with many concurrent users you want a large stripe depth so that a single I/O request can be statisified with just one disk/head. This allows the possibility that other drives in the set can satisfy I/O requests from other users.*
The SAME folks say to use a depth of 1M, regardless of the type of data requests. See http://www.oracle.com/technology/deploy/availability/pdf/oow2000_same.pdf
Their argument seems to be that for depths of under 1M the head seek time starts to become a significant part of the response time. For depths of over 1M the reduced amount of head seek time does not reduce the response time significantly. I don't buy their argument completely but it's probably good enough for the work-a-day DBA.