Presented by:  Jeremiah Wilton, Blue Gecko

Jeremiah’s patented Breaking Oracle seminar – a one-hour condensed version of a full-day class – was again a huge hit.  To a packed room, Jeremiah mercilessly rained spins, hangs, corruption, and a variety of other Oracle failures and disasters on a poor, unsuspecting test 11hg database, then demonstrated his methods for capturing and analyzing them.  

Again, I won’t re-create his presentation in this post, but I’ll upload the slides as soon as I get them (my bad – I was just there!) and we also caught the entire presentation on video.  

Until then, here are few photos...

Breaking Oracle - caught in the act.

Anwering attendee questions - every Oracle disaster story you can imagine:

Presented by:  Craig Shallahamer, Founder, Orapub

I attended Craig’s second presentation, this on In-Memory Undo (IMU).  This was another great presentation on a feature I have to admit I really knew nothing about.

In a nutshell, “traditional” undo uses physical blocks, just like any other segment.  If a query needs to read an undo block to maintain read-consistency, it still has to go through all the latch and potentially I/O overhead of any other query.  Additionally (and this was new for me, but now seems obvious), undo block changes also must maintain redo information for recovery purposes.

In-memory undo replaces undo blocks with an in-memory control structure called an IMU node.  Instead of writing a change to an undo buffer, undo is written to the IMU node and persists in memory even after commit.  A read-consistent query needs none of the overhead associated with physical blocks, so speed is dramatically improved.

Craig’s tests showed a 21% drop in CPU time and a 6.5% drop in wall-clock time for a load test that used IMU versus one that used physical undo.  He attributes the not-as-dramatic wall-clock improvement to I/O overhead that isn’t reflected in the CPU timing.

Overall, a great presentation on a feature I knew nothing about.  You can download the presentation and find more details at Orapub:  http://www.orapub.com

Presented by:  Jeremiah Wilton, Blue Gecko

OK, so, no, I’m not going to steal Jeremiah’s thunder here.  What I am going to do is post his presentation, and give you a teaser:  All bias aside, Real Application Testing is a truly revolutionary 11g feature that every DBA should get to know.  As Jeremiah demonstrates, its uses go far beyond planning an 11g upgrade.  In his presentation, he outlines a strategy (showing his work) to iterate over proposed parameter or system changes and measure the actual result before production implementation.  He compares his results to the AWR database advisors, and the results are very interesting.

Here’s the presentation – more on this later:

Deriving Optional Configuration Values Using 11g Database Replay

If you’re at Collaborate tomorrow (Wednesday 5/6) and you don’t catch Jeremiah’s next presentation, you’ve wasted your money.

As an Oracle Apps administrator, it's a good idea to have a punch list of things to check after your applications have been started, to ensure all services are working correctly.  For Release 11i, the following steps are a good starting point.  It will take a few extra minutes to work through this list, but it's certainly more aggravating to get the call from one of your best users relaying problems an hour after you've handed the system back

1. Verify you're not in maintenance mode by logging into sqlplus and running the following query

FND_PROFILE.VALUE('APPS_MAINTENANCE_MODE')
--------------------------------------------------------------------------------
NORMAL

If this returns 'MAINT', then you need to light adadmin up and get out of maintenance mode.

2. Check Apache & JServ log files for errors in startup process.

•  under $IAS_ORACLE_HOME/Apache/Apache/logs look for problems/errors since starting services back up
• the latest access_log_pls.* file
• the error_log file
• the latest access_log.* file
• check the $IAS_ORACLE_HOME/Apache/Jserv/logs/mod_jserv.log file for errors

• Open a form and query data (Security->User->Define or Application->Network Test for example), and ensure there are not any errors.
• Open the Administer Concurrent Managers Form (Concurrent->Manager->Administer) and make sure the Service Manager: <SID> is running.  If it's not, check the internal concurrent manager log file ($APPLCSF/$APPLLOG/<SID>_<MMDD>.mgr, where MMDD is the current month and day) for errors.
• Launch a harmless report, such as the Concurrent Program Details Report
      Application Name = Application Object Library
      Program = Concurrent Program Details Report
• In a perfect world, there would be a way to test a report that uses the OPP service.  One way to do this is to modify the concurrent program definition of the Concurrent Program Details Report to produce PDF output.  To do this pull up the definition in the Define Concurrent Programs Form (Concurrent->Program->Define), and under the output section,  change the Format value from 'Text' to 'PDF' and save your changes.  If you execute this successfully, then your OPP processor is working fine.
• After you've successfully executed these jobs, pull up your request history (Concurrent->Requests) and export the data (File (menu at the top of the applet - not in the responsibility menu)->Export.  You will probably be prompted that you have more than 100 records, and this record set is plenty for testing.  Press the Stop button.  Save the exported data and verify that the data looks correct.
  

4) Make sure the Self Service Products are working - it's common for the Oracle System Administrator to have iExpense or iProcurement, as they would naturally need to log expense reports or possibly make purchasing requests.  Simply change to one of these responsibilities and navigate around a bit to ensure there are no errors.

5) Check Jserv with the following url
    http://<hostname.domain>:port/servlets/Hello

 6) Log into the Oracle Application Manager (OAM) as the System Administrator

    http://<hostname.domain>:<port>/servlets/weboam/oam/oamLogin

•   Navigate to the Workflow Manager section, followed by the service components section.  Ensure that the Workflow Notification Mailer and Agent Listeners are running.  If this is a test environment, you may explicitly not want these up - especially if you've not set the WF Notification Mailer mailer override address (which prevents notifications from your test system going to real end users leading to confusion).

http://<hostname.domain>:port/OA_HTML/jsp/fnd/aoljtest.jsp

Presented by:  Adam Lee, Product Manager, Oracle

My first session of the day was a little dry.  The overview was very typical Oracle stuff:  A bunch of slides displaying block-pyramids with each block sporting important terms like “INFRASTRUCTURE” and “ABSTRACTION ENGINE.”  I was hoping to see potential support impact on Blue Gecko.  Eventually, some of our E-Business Suite customers, if not others, will implement BI.

Truthfully, I don’t know a tremendous amount about implementing Oracle BI.  This was an overview session that explained the high-level BI components and how they allow seamless heterogeneous access to heterogeneous data sources.  It all sounds fine, but when pressed with questions regarding configuration difficulties, Mr. Lee deflected a bit; this was not a technical session.

Without repeating the entire content of the presentation, the success of BI depends on configuring the Common Enterprise Information Model, which is a core function of the BI server itself.  Defining business rules, dimensions, access rules, etc. is the entire chore – if your data is very disparate and does not relate well across systems, this session implied you might not get a ton out of BI until that’s addressed.

The discussion degraded at one point into (substantial) confusion regarding licensing, discoverer, and other details.

The whole thing sounds very confusing – it might have been better to provide a demo with the overview.

On the other hand, there are packaged BI solutions for E-Business Suite and other Oracle applications (PeopleSoft, Seibel, etc.) with pre-built metadata.  I expect that the ramp up time on these is far shorter than, say, linking a bunch of Microsoft cubes, Oracle E-Business Suite, and a legacy DB2 database.

Overall, pretty dry, a little confusing, but I think I have a better idea of what BI architecture looks like, which is what I was hoping for.  

First, there aren’t that many people here.  I don’t know the exact count, but I just don’t see too many crowds anywhere.   Not a big surprise considering folks would have been signing up during the height of the economic hysteria, but kind of disappointing all the same.

Mostly I’m here to meet with a few existing customers, a few prospects, and hopefully generate a few new prospects.  I’m also here to record Jeremiah Wilton’s presentations so we can post them on our blog.  Should be fun.  :)

Next impression is that the Orlando Convention center is HUGE.  I had a little trouble interpreting the room numbers on the schedule and ended up at the wrong end of the facility for my first (intended) session.  It was a loooong walk back.  Maybe the size of this place makes the attendance look smaller, but I don’t think so.

I'm all coffee'd up after a long day of flying yesterday and a 3-hour time difference, so let's see how this goes.