Log Buffer #84: A Carnival of the Vanities for DBAs

Feb 15, 2008 / By David Edwards

Tags:

It’s the 84th blog-tacular edition of Log Buffer, the weekly review of database blogs!

We begin with some Oracle security news. A tutorial of Oracle’s on defending against SQL injection attacks gets a good review on Pete Finnigan’s Oracle security weblog. Pete writes, “This is a superb tutoral, well written and positioned just right. . . . [It] starts by explaining what SQL Injection is with some good flash examples and how to avoid SQL Injection. It includes first order and second order attacks and also discusses reducing the attack surface, removing API’s, use of invoker rights, reducing arbitary inputs and more. . . . This is a good document and one of the best security documents I have seen from Oracle.”

Systems Engineering and RDBMS reports on Sequence Enhancement in Oracle 11g: “Starting with Oracle 11g, we can use sequences with straight variable assignment. Before 11g, we always have to use SELECT INTO clause to get sequence value in the variable.” Examples provided.

Christian Bilien demonstrates that the “log file sync” wait event is not always spent waiting for an I/O, it may be trying to get some CPU attention.

On her Oracle blog, Girlgeek is trying to lose a datafile — on a Windows system in particular, but some the post also has some comparisons with this experiment on other OSs.

Jonathan Lewis shines a light on a little bug in index rebuilds in Oracle 10g.

Cary Millsap has the first part of the tale of how OFA (Oracle Flexible Architecture) began. It comes from him, and it started like this: “So I created myself a standard.” Marco Gralike responds with his own story about OFA and other standards, such as SAME.

While I’m talking about oaky people and their stories, here’s Jason Arneil offering his book review of Tales of the Oak Table, not a new book but arguably a classic.

The Ardent Performance Computing blog posts this howto on Oracle Clusterware on RHEL5/OEL5 with udev and multipath.

On Preferisco, Nigel Thomas rounds up some recent posts on schema version control and offers his own perspective. “Code control is a walk in the park compared to ‘schema control’. . . And ‘data control’ adds even more challenges.”

Nigel missed one post, though — Sheeri Cabral’s item on source controlling the database schema here on Pythian’s blog. Sheeri (a.k.a., the MySQL SheBA) also finally caves, and tells why Postgres is superior to MySQL. (While saying “Uncle!”, I assume.)

Reaching back a little into last week, MySQL AB’s (can I still call them that?) Jay Pipes had something to say about the sunny future the two prominent OSS DBMSs. Jay writes, “There were very few in the MySQL community who saw the [acquisition of MySQL by Sun] as a bad thing — most opinions seemed to show a keen upside to having the resources at Sun available to improve MySQL. There were, however, a number of folks in the PostgreSQL community who openly fretted about whether the move signaled the demise of PostgreSQL support at Sun. . . . Let me explain why I see the reverse being true.”

Kaj Arno, also of MySQL, has this revealing post about part of what Sun is taking on: MySQL culture, featuring a list of what MySQL employees believe that is — for example, “Little ‘red tape’”, and “FUN FUN FUN”. And the “no-@sshole rule”.

Frank Mash links to a Sun Microsystems study on MySQL and Solaris in his post, Scaling the third most active Social Network with MySQL/InnoDB/Solaris, which also serves as a preview of Frank’s presentation at the MySQL Conference 2008.

MCslp Coalface’s Martin Brown links to an older item of Frank’s and covers some detail about MySQL on OpenSolaris.

Johan Andersson, a high-availability consultant for MySQL, offers an expert look into the features of MYSQL Cluster.

On High Availability MySQL, Mark Callaghan delves into the use of ANALYZE TABLE on transient tables.

Moving to SQL Server blogs, MaasSql looks at why views are evil: “Synopsis: Views are evil, bad, buggy, temperamental, tortuous, and should be avoided.” It’s a little less black-and-white than that, of course.

Adam Machanic posts on solving the top-per-group problem. He moves from a note on developmental psychology to a discussion of the four basic approaches he has discovered. Very nice.

David Portas read something Bill Gates said about “weak data models,” and responds with some observations about strong model and weak tools.

Joseph Sack has put together an example of how SET options impact the Query Optimizer index choice.

Joe Webb published an excellent introduction answering the question, does the order of columns in an index matter? Give this one to your interns and newbies.

Evidence at hand suggests that SSIS may cause insanity and/or addiction. Perhaps severe headaches too. First case in point. Musings of an SSIS Madman has part-two of a series on their SSIS performance framework. Maybe not so much for newbies: “This is a medium intensity post (you should not get a migraine, but I make no promises).”

Next, Jamie Thomson, the SSIS Junkie, reports on case-sensitivity in the SSIS Lookup component. He writes, “[Jorg Klein] correctly observes that: A lookup transformation uses full caching by default. This means that the first thing it does on execution, is loading all the lookup data in its cache. When this is done it works as expected, but with case sensitivity.” A couple workarounds are on offer, as is the beginnings of a write-in feature campaign.

In PostgreSQL blogs, the Postgres OnLine Journal has an item on using DML to generate DDL and DCL. Sounds fancy doesn’t it? David Fetter thinks it’s “pretty Rube Goldberg”.

Jignesh Shah posts a HOWTO on using sysbench with PostgreSQL on Solaris.

Peter Eisentraut has a little postcard from the Prague PostgreSQL Developers’ Day 2008. He also writes that is was perhaps the first (PostgreSQL?) conference with its custom wine. Peter has a pic of a wine bottle with a little blue elephant on it. Excellent vintage, the 8.3!

With striking coincidence, Kristian Köhntopp introduces the world to Salmiakki – the official MySQL drink. Take a look and see if you don’t agree that Kristian has given new meaning to “mysqldump“.

That’s all. As always, your Log Buffer coordinator’s exhortation to get involved: submit your favourite blog items, or publish an edition yourself and make friends with a world of DB bloggers. Just send me an email to get started.

Until next time!

Leave a Reply

  • (will not be published)

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>