Actually a very good dasd benchmark tool to use is iozone, if anyone is interested
No doubt iozone does very extensive tests to show different type of cache and blocking. Often it is hard to tell what these results to do real life applications. Especially with something as complex as disk I/O. Also, when the benchmark does the measurements on Linux and Linux is not able to keep track of time (because it runs on z/VM) the values are not very useful (I have bonnie reports tell me it used 220% of one CPU). The Redbook referenced in the thread (which I recommend strongly ;-) compares Linux numbers with VM numbers to validate the results.
Another thing to keep in mind is that imho with Linux on z/VM you care about total progress made on the system, not necessarily the maximum throughput achieved in a single virtual machine. The is measuring the mileage of your car versus top speed. Most likely you don't achieve the best mileage operating at top speed.
Rob
