Log Buffer #31: A Carnival of the Vanities for DBAs

Welcome, to the 31^st edition of Log Buffer, the weekly survey of database blogs.

The DB2 world is a good place to begin. On Monday, Craig Mullins published a DB2 for z/OS status check, What’s Up With DB2 Today? on his Perspectives on Database Management: “I thought it might be useful to examine what versions are available and will soon be available, and what it might mean for DB2 users.” And then on Tuesday, IBM went and cheated the prophet by making its official announcement about those versions. Craig said, “Gee, If I Had Only Waited a Day.”

In short, DB2 V7 is fast approaching its official end-of-service. On Getting the Most out of DB2, Willie Favero gives some details. In another entry, he says, “I am really getting tired of people telling me how bad they think DB2 for z/OS Version 8 is,” and that V8 really is better than DB2 Version 7.

Another blogger trying to correct errors of one kind or another was Josh Berkus on Database Soup. He wants to set the record straight on PostgreSQL support in the wake of an interview in which he feels he was misrepresented.

IBM’s other DBMS, Informix, is also moving ahead. On Informix technology, Fernando Nunes mentions the the first public customer workshop on the next major release, code-named “Cheetah” (because it’s fast or because it’s spotty?).

Here’s another helpful item about Oracle and the DST change, from Chris Foot’s Oracle10g Blog. He mentions that Oracle now has a DST Support area in Metalink, and that a number of webcasts on the subject are on their way. Quote: “(We) need to ensure that our databases, administration toolsets and applications are 2007 DST compliant. But we also need to … ensure that all of the components (operating systems, third-party applications, middle-tier products, external Java engines, etc.) that interact with our databases are also 2007 DST ready.”

Is Oracle acting out of arrogance or altruism in the matter of their Unbreakable Linux venture? That’s the question posed on Eye on Oracle by Elisa Gabbert. She surveys a range of opinions, including that of an Oracle executive who concurs that Oracle’s motivation is selfish, and that that is, he says indistinguishable from altruism.

A lot of DBAs are tackling upgrades to Oracle 10, and here are a couple items on this. One post on the 10g upgrade is from Jonathan Lewis’s Oracle Scratchpad: “When you upgrade to 10g from 9i you may find lots of execution plans suddenly ‘go wrong’.”. Jonathan provides three reasons for this, one of which is a difference in dbms_stats packages that might render your old way of gathering stats useless. “Because of this particular change, if you did (sic) specify a method_opt in your scripts under 9i you were not generating histograms; under 10g you will be collecting histograms on any columns that Oracle thinks might be suitable candidates. The effect could be dire.”

The second is from Jeff Hunter on the So What Co-operative, who points out a Metalink note regarding upgrading to 64-bit 10.2.0.3.

Daniel Fink of OptimalDBA says SQL is a lot like Lego: “You don’t have a lot to work with, but a little imagination can lead you to write some amazing code.” On the AMIS Technology blog, Lucas Jellema provides a case in point with his piece on writing a pure SQL-based string tokenizer. Pure hacking for the fun (and the utility) of it. Hey, this must be where database administration and development meet.

Speaking of hacking, but in the other, bad, sense of the word, Brian Knight, SQL Server MVP links to a video presentation demonstrating how an SQL Server application can be hacked (or cracked) using those favourite old adversaries, SQL Injection and cross-site scripting.

On Kimberly L. Tripp‘s blog appeared an item on “the clustered index debate” in the SQL Server world. “In the years since the storage engine was re-architected (SQL Server 7.0+) there’s been constant debate on how to appropriately choose the clustered index for your tables. I’ve generally recommended an ever-increasing key to use as a clustered index and many find that counterintuitive.” Performance is the crux of the debate, which the article describes in detail.

Turning to another open source RDBMS, on IBDeveloper, Dmitri Kuzmenko answers the question, does Firebird run on Windows Vista. Not to give too much away, the answer is yes, and Dmitri gives details on his install of Firebird 2.01 RC1 on Vista Business.

Completing the picture, Firebird News announced on Thursday that Firebird 1.5.4 is released for Windows and Linux. “A number of additional retrospective fixes have been introduced for bugs that became apparent and were fixed in the Firebird 2 tree during the Firebird 2.0 beta cycle.”

A couple of Brian “Krow” Aker‘s Idle Thoughts on why details in MySQL do matter. First, connection-pooling: “Every so often I have someone ask me ‘does it really matter if I pool connections?’. The answer is yes!” Number two: paying attention to the open tables setting, and the significant difference it can make to write performance.

On Jeremy Cole’s weblog, he alerts readers to a MySQL bug involving corrupted relay logs. “It appears to have something to do with using BLOB or TEXT fields in replication. Are you seeing slaves stop with corrupted relay logs? Does restarting replication using CHANGE MASTER and the Exec_Master_Log_Pos from the stopped slave1 work just fine? Do the masters binary logs look perfectly OK?”

On TheOpenForce.com, Zack Urlocker reports on two new MySQL storage engines: Falcon and PBXT. I have to sheepishly admit that I had never heard of PBXT before now. Both of these engines offer transactional functionality, and they have both been recently updated. Zack has a set of links apropos both of them.

An item called A ‘Simple’ Protocol for Manual MySQL Slave Promotion to Master appeared on Kevin Burton‘s Feed Blog. “(If) a MySQL master fails, most people just deal with a temporary outage. They bring the box back up, run REPAIR TABLEs if necessary, and generally take a few hours of downtime. … One solution would be to use a system like DRDB to get a synchronous copy of the data into a backup DB. … You could also use a second master in multi-master replication… A simpler approach is to just take a slave and promote it to the master. If this were possible you’d be able to start writing to the new master almost immediately after the old master fails.” Kevin lays out the required steps to take.

I’ll make my exit here. Log Buffer is in need of some new blood, so if you’d like to edit and publish an edition of your own, take a look at the Log Buffer homepage to learn more about it.

Until next time!