RE: EBS (12.1.3) / HCC and SSC
- From: "Mark W. Farnham" <mwf@xxxxxxxx>
- To: <niall.litchfield@xxxxxxxxx>
- Date: Wed, 7 May 2014 18:06:22 -0400
That is correct. You’ll want to be sure the exact patch level of the Exadata
box you are using is on the EBS certification matrix for the release and patch
level of EBS you are using.
That makes you certified and eligible for support if the appropriate contract
is paid up.
If your HCC usage triggered a bug in HCC, I suppose undoing the HCC might be
the recommended fix. But without a bug, the EBS application should not be able
to tell the difference, and it would be a bug if it could. But you’re right
that it might be an HCC bug. In fact if the EBS applications can tell the
difference I think that defines it as an HCC bug. Then the question would be
whether or not to hold your breath until a patch was made to the hypothetical
HCC bug. I’d recommend against any such breath holding.
AND they do not recommend it for performance reasons due to their testing.
In the case where the OP was going to do some sort of time based partitioning
and HCC only the old stuff, it seems highly unlikely there would be a problem.
Conversely it would probably be brain dead to HCC current invoices on and do a
check run.
mwf
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On
Behalf Of Niall Litchfield
Sent: Wednesday, May 07, 2014 4:58 PM
To: Mark W. Farnham
Cc: ORACLE-L; Frits Hoogland; George.Leonard@xxxxxxxxx; jpiwowar@xxxxxxxxx
Subject: RE: EBS (12.1.3) / HCC and SSC
Right but EBS at least used to be really grotty about 'support' vs certified.
Things may have changed but back in the 11.5.10 time frames I'd expect any
support ticket to end with can you reproduce on a system without HCC if the
diagnosis dumps showed HCC in use. Good luck with that.
On 7 May 2014 17:44, "Mark W. Farnham" <mwf@xxxxxxxx> wrote:
Frits is basically correct, as confirmed by ATG below in response to my request
for clarification:
Am I wrong that the applications could not really know whether the physical
storage is using HCC by the time it accesses the data?
I wonder whether this is failing to understand the difference between “your
support does not include Oracle evaluating and setting up HCC for your
particular system implementation and if you screw that up backing it out is
your problem” and “if you use HCC on any part of an EBS database your
installation is unsupported” and the huge range of possibilities in between.
Mark,
Thanks for the email. You’re not wrong, an EBS instance would not know if HCC
is in use. To clarify the recommendation: The Oracle E-Business Suite
Performance Team does not currently recommend HCC for EBS. Our testing has
shown degradation in performance with HCC and EBS (an OLTP system with
high-volume tables).
Thanks.
~ep
Elke Phelps | Product Management | 303.661.2475 |
<mailto:kenneth.baxter@xxxxxxxxxx> elke.phelps@xxxxxxxxxx
Oracle E-Business Applications Technology Group
ORACLE | 600 Oracle Parkway | Redwood Shores, CA 94065
<http://blogs.oracle.com/stevenchan> Oracle E-Business Suite Technology Blog
Hardware and Software Engineered to Work Together
From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On
Behalf Of John Piwowar
Sent: Wednesday, May 07, 2014 11:26 AM
To: Frits Hoogland
Cc: George.Leonard@xxxxxxxxx; ORACLE-L
Subject: Re: EBS (12.1.3) / HCC and SSC
Understood. There are also patches which perform mass data fixes/updates,
though, and those could be more problematic in the "HCC doesn't like changes"
context.
The real reason for lack of support (or lack of documented support), however,
could be more prosaic: number of EBS customers on Exa platforms. Given the
range of customers they support, and the number of moving parts in EBS, the ATG
team has to be selective in the database features that they test and certify.
That list, as I understand it, can be heavily influenced by customer feedback.
If enough customers ask for a statement of support for HCC in EBS, the ATG team
might take a look; they tend to be very responsive to that sort of stuff.
Please note that I don't have any inside info on the inner workings of the
Oracle EBS ATG. This is just a summary of my impressions from years of reading
Steven Chan's blog, and should not be misinterpreted as an official position on
*anything.* ;)
On Wed, May 7, 2014 at 7:53 AM, Frits Hoogland <frits.hoogland@xxxxxxxxx> wrote:
Please mind the partitioning scheme and the table fields are not changed in any
way with HCC, it's the way the data is stored which is only changed. So
subsequent patches which alter table or partitioning properties can happily do
so. It would have an effect on the compression (effectivity), but that's all.
Indeed, I think HCC should be seen as 'yet another way to compress'.
frits
On May 7, 2014, at 4:46 PM, John Piwowar wrote:
I think the sticking point is here:
> So if you've got partitions with data that will not change anymore (HCC
> compression doens't like changes), there is no reason not to compress it.
It's not unusual for an EBS patch to alter the data or structure of a table or
its dependent indexes. This is true even for some of the Very Important Tables
in AR, Order Entry, etc., and particularly true during upgrades. So, just as
you would need to be careful when implementing a custom partitioning scheme
(supported, but with similar caveats), you might run into maintenance issues
for tables to which HCC is applied.
The ATG team has done extensive testing with the Advanced Compression (compress
for OLTP, formerly "for all operations") feature, and has published a white
paper on which objects were good targets for compression in both a test
instance and Oracle's production GSI. Obviously Advanced Compression doesn't
"squeeze the air out" as effectively as HCC, but it's an option.
Here's a link to the Advanced Compression discussion, with ref to the
whitepaper. It's from 2010, but some of the references are more up-to-date:
https://blogs.oracle.com/stevenChan/entry/new_whitepaper_advanced_compression_11gr1_benchmar
On Wed, May 7, 2014 at 6:01 AM, Frits Hoogland <frits.hoogland@xxxxxxxxx> wrote:
Hi George.
I am by no means an EBS specialist. I am aware a lot of database stuff needs to
be done in an 'EBS way'.
However. Essentially EBS is a client of the database (only oracle itself made
it, instead of a 3rd party)
HCC is a compression technique which requires no client SQL change to be used.
So if you've got partitions with data that will not change anymore (HCC
compression doens't like changes), there is no reason not to compress it.
Said that, you should be aware of anything that will break when doing the
mandatory recreation via direct path. (alter table move for example)
Frits Hoogland
http://fritshoogland.wordpress.com <http://fritshoogland.wordpress.com/>
frits.hoogland@xxxxxxxxx
Phone: +31 20 8946342 <tel:%2B31%2020%208946342>
Mobile: +31 6 14180860 <tel:%2B31%206%2014180860>
On 07 May 2014, at 07:41, George Leonard - Business Connexion
<George.Leonard@xxxxxxxxx> wrote:
<RSImage.jpeg>
Hi all
Ok, was a bit taken aback yesterday when an engineered informed me that HCC is
not supported on EBS (12.1.3).
I can't imagine that a Exa feature is not supported by now, nearly 6 years
after V1.
Is this really so, the client have data spanning 15+ years in their line items
tables, we want to partition the table and HCC compress OLLDDDD data. It was a
big part of the business case.
Luckily the SSC will be a consolidation platform sot he value of HCC is still
valid for the DW and other systems.
But really, Oracle not supporting HCC on EBS on Engineered systems... Thats
like something I would expect from SAP.
Please help - URGENT. (If you have a note number that tells otherwise, and
specifies how to partition EBS would appreciate it)
(ME <> EBS expert).
Yours Sincerely
________________________________________
George Leonard
Oracle Engineered System Specialist
Mobile: +27.82 655 2466 <tel:%2B27.82%20655%202466>
eMail: george.leonard@xxxxxxxxx <mailto:george.leonard@xxxxxxxxxx>
Web: <mailto:george.leonard@xxxxxxxxxx> http://www.bcx.com
<image003.gif>
Other related posts:
