On 12/18/09 10:47 AM, "Szefler Jakub - Hurt TP"
<[email protected]> wrote:

> Hi,
> Why ? 

Mainly because the Clariion boxes are designed for low-cost, medium-to-low
performance storage uses and putting z/VM data on them tends to push their
design limits very, very hard. It technically works (once you figure out
some of the occasionally bizarre WWPN/masking tricks that they needed to
become visible and work, and how to get z/VM to do multipathing with them),
but we saw some unexplainable delays in reading and writing blocks when
being pushed very hard (eg, using them for VM paging disk or EDEVs with
heavily used minidisks).

When we bugged EMC about it, they weren't very comfy with helping us to find
the problem, citing the same answer wrt to the DMX storage you mentioned.
The discussion wrapped up with a reference to the above as that EMC
suggested using them for applications that didn't push the Clariion boxes so
hard, eg scratch space and/or directly-attached Linux data partitions that
behaved more like traditional discrete server usage of the Clariion boxes.

Sounds like they've spent some time fiddling with the newer boxes to get
them working better. Also the recent Clariion boxes have gotten significant
upgrades in controller hardware, so maybe some of the problems we had have
been mitigated, but I haven't gone back to check that.  

----------------------------------------------------------------------
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

Reply via email to