Pythian Goodies - PX

Doug's Oracle Blog

  • Home
  • Papers
  • Books
  • C.V.
  • Fun
  • Oracle Blog
  • Personal Blog

Feb 27: Pythian Goodies - PX

When I was in Ottawa, Christo Kutrovsky asked me to talk about Parallel Execution for the series of Pythian Goodies he's organising. After I'd done it, I insisted that he let me see the video before it was published on the web so that I could make sure that I was happy with it.

After watching it (which is a tough call in the first place - you should try it!) I had a page full of notes of things I'd said unclearly, misleadingly and some of doubtful accuracy too. There are times when I wonder about the rubbish that can tumble out of my mouth for a few moments. Most instructors will tell you that they've experienced that more than once - mouth open before head fully into gear.

But, rather than whine about it and refuse to let it out, I thought it would be okay because

1) I don't underestimate how much some people might want to see it. Not so much for the technical content as to "meet" me ;-) That's not a conceited as it sounds because I've experienced for myself what it's like to meet someone who you only know electronically, and it's a super-duper version of 'putting a face to the name'. With the video out there, I can safely stick to photos of The Boys (and Polly) for my blog. (Oh, and I do recognise 'me' in this video, at least I got that bit right ;-))

2) Christo was very specific in his intention that this would not be a presentation, but an informal discussion (note the use of the term 'pissed-off users'!), and so he wanted me to talk off the top of my head about some things I thought a wide range of DBAs with different skill levels and interests might find useful. He did not want me to repeat any presentation that I might have spent a lot of time preparing. Christo talks about the format more in the announcement blog. I think I got the general idea but as close friends won't be surprised to hear, the audience didn't get as much of a word in as they're supposed to ;-)

3) My preparation time immediately before I stood up was filled with answering an urgent customer email straight before walking in to start, so I was a little flustered to say the least!

I promised myself I would limit the excuses and self-flagellation, so I'll leave it there. I still have that list of slightly stupid things I said but I'll be keeping it to myself until I see who picks up on what ;-) I might post some basic corrections later.

Keep an eye out for it on Pythian's blog because I think it will be appearing soon, maybe later today.

The bit you'll miss, sadly, is the post meeting discussion over pizza :-(

The bit you'll miss, hopefully, is the fact that my flies are undone for the entire presentation!

The bit that I hope some well-known Oracle types might miss is me ripping them off, arguing with their views and name-dropping shamelessly! I must have forgotten the video camera and I seem to recall I make reference to this towards the end!

Let me know what you think of it, anyway, but only if you're nice about it ;-)
and you should feel more than free to post questions, corrections and so on to the blog itself, so that they exist alongside the video.

Posted by Doug Burns Comments: (12) Trackbacks: (0)

Trackbacks
Trackback specific URI for this entry

No Trackbacks

Comments
Display comments as (Linear | Threaded)

#1 - Alex Gorbachev said:
2007-02-27 17:23 - (Reply)

The bit you'll miss, hopefully,...[/i]
Not anymore!!!

#2 - Doug Burns said:
2007-02-27 22:50 - (Reply)

I was informed by a friend this evening that ....

I am sure that you tried to tempt viewers by the prospect of pornography and freak trouser accidents - but on my display in the office I could see nothing amiss. Well apart from some presenter disappearing from shot at whim - I would not be surprised to see him exiting on the left with a plank only to reappear on the right at the other end of the same plank.

I can only apologise to those who I'm sure tuned in for the freak trouser accidents, but it would appear that Christo (rather generously) trimmed the bottom (so to speak) off the video.

I've just watched the original and, whilst it's not offensive, it's clear that my flies are, in fact, undone.

I was hoping it might distract from the content. Oh, well ...

#3 - Dominic said:
2007-02-28 12:26 - (Reply)

It's probably a problem at my end, but it keeps stopping at about 1/3 of the way through. But from what I'm watched - excellent.

#3.1 - Doug Burns said:
2007-02-28 13:49 - (Reply)

Thanks Dominic, I'll pass on details of your problem.

#3.1.1 - Doug Burns said:
2007-02-28 13:54 - (Reply)

Perhaps 'the problem' might be a better description for now.

#4 - Patty C. 2007-02-28 14:38 - (Reply)

If I have time before the weekend I will add it to my IPOD Video so that we can all watch it in the hotel bar at the Symposium next week. :-)

#4.1 - Doug Burns said:
2007-02-28 15:06 - (Reply)

No!

You're *too* cruel ;-)

#5 - Don Seiler said:
2007-02-28 17:45 - (Reply)

Doug,

Excellent presentation. I immediately went to check at my own database and saw that a user's ad-hoc query was running 8 slaves. I then shrieked in horror when I saw all the tables with DEGREE='DEFAULT' in DBA_TABLES. I'll have to fix that.

One issue is that I find the video lagging behind the audio further and further as it goes on. By the end it's almost 2 seconds off.

Also I'm posting here since the pythian site seems to be down at the moment.

#5.1 - Paul Vallee said:
2007-02-28 20:59 - (Reply)

Hi Don,

>Also I'm posting here since the pythian site seems to be down at the moment.

You're right. It's back now, sorry about that. We had an issue with the linux kernel on the database server while doing some clearly too-ambitious maintenance work mid-day. It's fixed now.

Cheers,
Paul

#5.2 - Doug Burns said:
2007-02-28 23:06 - (Reply)

Thanks a lot Don, that's good to hear. I'll blog about the lag and the outage tomorrow ...

#6 - Don Seiler said:
2007-03-01 03:12 - (Reply)

This might be a dumb question. If none of my objects have parallel on them (i.e. dba_tables.degree='1'), will supply a parallel query hint apply to both the query *and* the index usage (if the CBO chooses that route)?

#6.1 - Doug Burns said:
2007-03-01 10:11 - (Reply)

Don,

I don't think it's a dumb question at all.

My first reaction would be - I'm not sure. I could guess, but, taking a tip I've heard Jonathan Lewis make a number of times, more hints and more explicit hints are likely to get you closest to what you're looking for.

Based on that, my recommendation would be that if you want a parallel index scan, I wouldn't depend on the /*+ parallel */ hint, but would include an explicit /*+ parallel_index */ hint.

Some might say that you shouldn't be using hints at all but, if you already intend to hint for the table scan, I would also hint for the index range scan.

I'd also suggest that the best way of proving this would be to test it against your data, your indexes and your queries. That's not a cop-out, it's just more likely to give you results that are useful to you.

Hopefully you'll forgive a less than brilliant answer. I was going to set up a test to see, but I'm feeling pretty under the weather at the moment and have a mountain of stuff to get through.

askdoug.com's never going to take off, it it?! ;-)

Cheers,

Doug

(Oh, and the sooner the JL CBO volume covering parallel comes out, the better, but I think that's a while away yet.)


Add Comment

Standard emoticons like :-) and ;-) are converted to images.
E-Mail addresses will not be displayed and will only be used for E-Mail notifications.
BBCode format allowed
 
 

Statistics on Partitioned Tables

Contents

Part 1 - Default options - GLOBAL AND PARTITION
Part 2 - Estimated Global Stats
Part 3 - Stats Aggregation Problems I
Part 4 - Stats Aggregation Problems II
Part 5 - Minimal Stats Aggregation
Part 6a - COPY_TABLE_STATS - Intro
Part 6b - COPY_TABLE_STATS - Mistakes
Part 6c - COPY_TABLE_STATS - Bugs and Patches
Part 6d - COPY_TABLE_STATS - A Light-bulb Moment
Part 6e - COPY_TABLE_STATS - Bug 10268597

Comments

Doug Burns about 10053 Trace Files - Different Plan in Different Environments
Tue, 02.04.2013 08:57
You're welcome. Now I just nee d to pull my finger out and ac tually come up [...]
Howard Rogers about 10053 Trace Files - Different Plan in Different Environments
Mon, 01.04.2013 23:08
Makes a big difference, so tha nks for that! With two brow ser windows, o [...]
stelioscharalambides.com about 10053 Trace Files
Sat, 30.03.2013 16:28

Upcoming Presentations

Bookmark

Open All | Close All

Syndicate This Blog

  • XML RSS 2.0 feed
  • ATOM/XML ATOM 1.0 feed
  • XML RSS 2.0 Comments
  • Feedburner Feed

Powered by

Serendipity PHP Weblog

Show tagged entries

xml 11g
xml ACE
xml adaptive thresholds
xml ASH
xml Audit Vault
xml AWR
xml Blogging
xml conferences
xml Cuddly Toys
xml Database Refresh
xml DBMS_STATS
xml Direct Path Reads
xml Fun
xml grid control
xml hotsos 2010
xml listener
xml Locking
xml oow
xml oow2009
xml optimiser
xml OTN
xml Parallel
xml Partitions
xml Patching
xml swingbench
xml The Reality Gap
xml time matters
xml ukoug
xml ukoug2009
xml Unix/Shell
xml Useful Links

Disclaimer

For the avoidance of any doubt, all views expressed here are my own and not those of past or current employers, clients, friends, Oracle Corporation, my Mum or, indeed, Flatcat. If you want to sue someone, I suggest you pick on Tigger, but I hope you have a good lawyer. Frankly, I doubt any of the former agree with my views or would want to be associated with them in any way.

Design by Andreas Viklund | Conversion to s9y by Carl