Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Usenet -> c.d.o.server -> Re: Is anyone else using 9i RAC on HP-UX ???

Re: Is anyone else using 9i RAC on HP-UX ???

From: Jonathan Lewis <>
Date: Sat, 19 Feb 2005 18:36:15 +0000 (UTC)
Message-ID: <cv80uv$9e6$>

but raised more questions than it answered.

The first big drawback, of course, was that their overview suggests that they used about 6TB of disc space for a 230GB database.

On it's own, that wouldn't really be a big deal - but I couldn't work out where the missing time was in the numbers produced, and if the missing time is related to disc response times, then the sparse use of discs becomes significant.

For Example:

  1. "Everyone knows" that using Async I/O increases the use of CPU, because of the hand-off and polling. (The first two words are in quotes, because possibly the statement is, nowadays, not true).
  2. It takes CPU to execute a transaction.

Under the medium workload on RAW, CPU utilization was 80%, at a transaction throughput (rate?) of 23,135. With the same workload and configuration, but using the Async I/O driver, the transaction throughput was 31,961 with a CPU utilization of 79%.

So where was the extra CPU going when the system was doing LESS work, and NOT handing off to the async processes ?

Example 2:

Why were the log file parallel writes 1-2 ms for the file system, but 4-8ms for the raw device ? It's not due to the file system cache, as the log writer is supposed to use write-through- cache to ensure correct protection of data.

There were a few other questions that came to mind as I read the figures. Mostly because I couldn't figure out why the numbers looked the way they did. Ideally, I guess, I wanted to see several other statistics too, such as latch spin figures, latch wait times, free buffer wait times, dbwr parallel write times, write complete wait times, log buffer wait times, and maybe a couple more.

Bottom line: it doesn't look like marketing, so I don't think anyone has tried hard to fake anything. But the figures look surprisingly good for raw, and I'd like to know if there was some quirk of the tests that just happened to work well with raw.


Jonathan Lewis
The Co-operative Oracle Users' FAQ
Public Appearances - schedule updated Jan 21st 2005

"hpuxrac" <> wrote in message

> snip
> RAW works fine on modern hpux but there is a need for additional
> knowledge on the part of the dba and more planning up-front.
> HP and oracle put out a paper not long ago claiming substantial
> performance benefits from using RAW and noted that many organizations
> will be converting over.
> Interesting eh?
Received on Sat Feb 19 2005 - 12:36:15 CST

Original text of this message