May 18: Log Buffer #45
I'll kick off this week with a discussion of the MySQL PBXT Storage Engine over on the MySQL Performance Blog. I find the availability of different storage engines one of the more fascinating aspects of MySQL, though I have to admit that when I start reading the functional details I tend to break out in a cold sweat.
"PBXT does not offer Durable transactions (meaning you can lose committed transactions if power goes down) furthermore in case of crash database may become corrupted, just as with MyISAM, and you might need to repair tables."
Still, as I keep being told; 'it's just a website, who cares if you lose a couple of transactions?'
Then again, software is rarely completely reliable. How's this for a bizarre bug in some versions of Oracle? Thanks to Peter K for that one.
Another old hand (he says, at the risk of upsetting both Peter and Noons) illustrates the all-round skills and imagination a DBA needs in a typical scenario of working with multiple versions and different problems. I always enjoy Nuno's blog - he has a wealth of experience, a distinctive voice and takes a mean photograph, too.
Actually, I have some sympathy with Jay Jannsen's point of view that MySQL is to Oracle as vi is to emacs or, to put it another way ...
"I'm simply saying that these technologies seem to represent a particular way of looking at developing software over time."
However, that doesn't really explain why, in editors, web development tools and any personal software, I desire the simplest possible solution (e.g. I'm quite happy with vi), but in the enterprise software I use at work, I want flexibility, which implies at least some complexity. Jay wraps up by asking of different technology geeks - 'Can't we all just get along?'.
Erm, no
As someone who has a slightly sad obsession with database server configuration standards, I noted Jeremy Schneider's posting about the naming of Oracle's redo log files, for which most people use a .log extension. I suppose I could imagine that sys admins are likely to view a .log file as a prime candidate for deletion, but isn't the underlying problem here to do with documentation, privileges, training and file locations? I can't say I've ever thought of it as a big risk, and I don't recall anyone deleting them because they looked like output logs, but I suppose it doesn't hurt to use .rdo and .arc for the archived versions, just to be sure. In any case, I really like the look of Jeremy's blog, which was a new discovery for me this week.
Back to MySQL again and the xaprb blog has a useful post discussing an approach to avoid problems with MySQL statement-based replication caused by temporary tables
DB2 next, with a review of the IDUG (International DB2 User Group) North American conference, courtesy of scotdb.
On to a subject that might seem a little dull to some but is central to the most successful database implementations - data integrity constraints. Craig Mullins is asking people to participate in a survey on database constraints over on his blog. It looks like the enquiry extends beyond the basic primary, unique and foreign key constraints, too.
Speaking of constraints, I'm not sure about this one from the Decipher Info Systems blog! Enforcing uniqueness across two tables suggests to me that there's something wrong with the database design and for further discussion and alternative views, see this thread at Tom Kyte's asktom.oracle.com.Minor insanity, of the best kind, from Lucas Jellema on the Amis blog. Remember when computing used to be *fun*? Oh, and once the monster is released from it's cage, there's no stopping it! Look, David Moloney over at SQLTeam.com is even starting on Chess!
Moving on to more practical SQL work over at Howard Rogers' blog, where he discusses the workings and importance of the MINUS set operator.
Sticking with SQL, here are are a couple of puzzles for you - the first is courtesy of Dominic Delmolino. Dominic's blog is always a source of wisdom, regardless of the particular DBMS you use. The next is from thecodist{}, with a response on Jeff's SQL Server Blog.
Laurent Schneider has a brief overview of the Oracle MERGE statement. What I particularly like about it is that, unusually, he shows examples that only perform an INSERT or only perform an UPDATE and shows how the syntax can help clarify conditional examples of those actions.
Oh, look! Check this out! Do you remember Database Appliances?
To finish off, I'm going to go all Web 2.0 on you - just to prove I can
First, Paul Gallagher has an interesting review of Oracle Technology Network's Semantic Web offering - an alternative source of weekly blog postings if your interests are limited to Oracle-related blogs.
Archive Log
And now for something completely different. I've had some recent communication with the Techdirt Insight Community, which is an attempt to build a more formal (and paid) dialogue between business and bloggers. Take a look at one of the open 'challenge issues' of the moment. As a died-in-the-wool enterprise DBA, Web 2.0 developments don't light my flame, but who's to say that one of these online database builders isn't going to be just what smaller companies are looking for? Hey, you don't get much more Web 2.0 than that! Take a look.Now it's time I went back to working on my presentation for the Miracle Scotland Database Forum (MSDBF). I'm looking forward to it, although I don't think Oracle guys are going to win any beauty contests!
Finally, in the spirit of platform-neutral DBA friendship, I leave you with - Pizza Beer!
#1 - Niall Litchfield said:
2007-05-18 19:16 - (Reply)
well I did take a look at the open challenge. please register it said.
nothing like this web 2.0 stuff for engaging the community eh?
#2 - Peter Lewis 2007-05-18 21:23 - (Reply)
I guess lager boys are allowed to hate real ale... but promoting pizza beer? Sorry, but no ![]()

