On 9/13/06, Pieter Harder <[EMAIL PROTECTED]> wrote:
I have one problem with the info provided by this reference.
Nowhere I can find does it talk about the dedicated FCP being single-path or
multi-path.
If the measurements were done with a multi-path setup --> great.
If the measurements were done with a single-path setup --> worthless.
Nobody is going to run single-path if they can avoid it.
Remember running dedicated FCP multi-path is done in software within Linux.
This is going to take cycles and skew the picture. Ditto for performance
instrumentation etc.
If this is not factored in the FCP versus ECKD picture is an apples to oranges
comparison.
Best regards,
Pieter Harder
[EMAIL PROTECTED]
tel +31-73-6837133 / +31-6-47272537
>>> [EMAIL PROTECTED] 09/12/06 8:47 >>>
For published benchmarks see "Linux Disk I/O Alternatives"
http://www.vm.ibm.com/perf/reports/zvm/html/520lxd.html
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
This beg the question, when is it advisable to use ECKD and when FCP?
From my own experience, I have run into trouble when Linux handles the
I/O's. I know prefer ECKD / mini-Disk for system related files.
I do have some more inexpensive disk "Clarion" that I set up with FCP
for testing purposes when reliability is not that important.
Everything else is on Symm with ECKD mini-Disks. We run SLES 9.3.
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
