Wednesday, May 24, 2017

It’s the end for sec_case_sensitive_logon

I’ve been using sec_case_sensitive_logon set to FALSE for quite a while. I do this because many of the scripts I use have been written over the year and the password in them has not always been well managed so to say.  However in 12.2 this has come to a hard stop.  Oracle has been saying this is being deprecated/unsupported and it sure is now!

If you set sec_case_sensitive_logon to FALSE in 12.2 it effectively makes sets the instance to “restricted logins” the only way to connect is with “AS SYSDBA”.   The weird thing is that the error you get doesn’t lead you do this at all.  So with is set to FALSE I’d get this behavior:

SQL> conn op/op
ORA-01017: invalid username/password; logon denied
Warning: You are no longer connected to ORACLE.

With it set to TRUE:

SQL> conn op/op

It sure would have been nice for something else to be mentioned about this.  I looked in the alert log and nothing there.   None of the trace files mentioned anything useful.  I search OTN and everywhere else.  Fortunately with some help from my buddies via twitter we were able to figure it out.  Thanks Jeremy!

Monday, March 6, 2017

Hotsos Symposium 2017 #hotsym17

The Hotsos Symposium this year was incredible.  It seems impossible but each year is better than the last.The mix of speakers and topics were excellent.  For me, it’s hard to pick a favorite, so many of the session were spot on.  Larry Carpenter opened with a great talk on Data Guard in the cloud, relieving some fears of many who worried that this would be a new leaning curve.  He showed that for the most part “nothing changes”.

Then Richard Foote took the stage to talk about new indexing features in 12.2, which you would expect from Mr. Oracle Index!  And the most important change is now we can have identifiers up to 128 characters so you can now finally create that index named:

That you always wanted too!  Of course this was in jest, it is cool to have longer names, but that isn’t really the best change. Much more useful are the new compression and the new index monitoring which finally really tells us what is going on with our indexes. 
After that things moved into multisession mode and folks have to make tough decisions on which  speaker to hear,  would it be Roy Swonger on performance stability after and upgrade, or Tim Gorman on Data Virtualization?   Or how about Mauro Pagano talking about SQL Plan directives or Stafan Koehler discussing Troubleshooting performance beyond wait events? Decisions, decisions.

Faster than I would think possible, it was Tuesday morning and Richard took center stage to talk about using AWR to help solve performance issues. Oddly he hardly mentioned indexing in this talk. He did show us how relaxing it is to be an Oracle professional in Australia, just hanging out listening to David Bowie and reading AWR reports.

Then back to multisession mode for much of the day and rounding out with Kerry Osborne talking BIG DATA. Interesting discussion on how all these pieces fit together, big data, transactional data, Oracle and Hadoop. There are a lot of moving parts going on here, it will be fascinating to see how this all works out in the coming years.

Now it was the Dinner and celebration with dancing video games, janga, foosball, air hockey and lots of geek talk of course!
 Wednesday morning was right into multisession mode with great sessions by Roger Cornejo showing customized database tuning tools to see beyond AWR, and Carlos talking SQL Trace TKPROF and plans for beginners.  Before lunch Toon Koppelaars and Bryn Llewellyn compared the “NoPLSQL” and Thick Database paradigms.  It was pretty clear that ThickDB was the winner in performance more often than not.  It's YUGE! 
And to finish the day we had a lively expert panel of Richard, Bryn, Kerry and Toon. Good questions from the audience on a wide range of topics.

  Training day on Thursday was nothing short of spectacular with Toon Koppelaars and Vlado Barun talking about real world performance with the driving message for us to use the Oracle database as it’s designed to be used.  Again the concept of ThickDB was key to the discussion.  They used a real time database to show how some changed helped and some didn’t.  Very enlightening to see this changed implemented and the dramatic affect they could have.   They showed getting 10X 100X and even 1000X performance improvements.   My big take away (as likely was for many others) is that connection pools are almost always sized poorly and mostly to big, pushing waiting to the client end showed YUGE performance gains.  And of course:


Wow what a week, as normal we all felt a bit off balance as our brains tried to assimilate all the great knowledge from the week.  If you were there you know the feeling.  If you weren’t, you need to be there next year.  The dates aren’t set just yet, but it’s always right around the beginning of March, be there and bring a friend!