Walden,
We have always assumed in the single level storage concept, that all
storage devices operate at the same approximate speed. Which for the most
part has been the case for some years. I think even when we could mix 10K
and 15K RPM disks in a storage pool, the difference wasn't as great as it is
with SSD and HDD. But remember we learned in middle school by George Orwell
that not all pigs are created equal. So, this the same for storage devices.
We still have a gigantic memory space, but some of the storage locations can
retrieve and write their data faster than others. All the OS is doing is
moving objects to the faster storage devices, and not breaking the single
level storage concept. It just continues to amaze me how far ahead of its
time OS/400 was, and how IBM just screwed up. But that is ANOTHER Thread,
that we have already beaten to death.
There are some even cooler optimization functions coming in the next
release of IBM i.
The question I was wondering last week, was how soon do you think it
will be until all our disk space is Solid State? Are we 3,5, or 10 years
away? I think it will be driven now by price, but when do they become
affordable? Look at TVs, you only see flat screen TVs when you go into the
electronics stores.
Pete
Pete Massiello
iTech Solutions
http://www.itechsol.com
Add iTech Solutions on Facebook:
http://www.facebook.com/group.php?gid=126431824120
Add iTech Solutionw on LinkedIn:
http://www.linkedin.com/groups?gid=2206093
-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Walden H. Leverich
Sent: Tuesday, September 29, 2009 9:08 AM
To: Midrange Systems Technical Discussion
Subject: RE: can POWER6 IBM i or System i run on non-IBM disk drives?
The concept of balancing data access for performance seems to fly in the
face of single-level-store, no? Isn't a core concept behind SLS that the
"higher-level" OS functions don't care about where on disk something is
stored, it's all presented to anything above the TIMI as one gigantic memory
space. If decisions on where to write data are being made based on knowledge
of the underlying storage infrastructure haven't we poked some rather major
holes in the MI layer?
-Walden
midrange.com
