Welcome to the 90th edition of Log Buffer, the weekly review of database blogs.
First, SSQA.net’s SQL Master offers his walk-through of best practices for installing SQL Server 2005, with clustering as the destination.
If you read SQL Server blogs, you already know Adam Machanic. I’m very pleased to mention his first post for the Pythian Group blog, covering the basics of minimal logging and its enhancements in SQL Server 2008.
Also looking at Katmai was Bob Beauchemin, with his tip on accessing multiple servers with the SQL Server 2008 PowerShell provider, something right out in the open that nonetheless you might have missed.
Bob also figures out a little more about 2008’s new sparse columns and column_sets.
Joe Webb’s site mentions his appearance on Buck Woody’s Real World DBA podcast, where they tackle the question, does JOIN order matter?
On OraStory appears a very-commented post by Dominic Brooks, tantalizingly called, The dea(r)th of Oracle RDBMS and contracting?. From the piece: “I feel like the war has been lost and there are only a few pockets of resistance left now, resistance that will sooner or later be squashed. The database is under attack. . . . A newly created hierarchy have decreed that databases are indeed bad. . . . And I was speaking to a friend today at a previous employer, a major media / entertainment company. They are planning to abandon their pragmatic approach to Oracle and switch wholely [sic] to open source databases, ORM tools, and the like.”
And speaking of “Oracle versus X” (why doesn’t HTML have a <segue> tag?) — in last week’s LB#89, Shakir Sadikali criticized a post by Sean McCown’s Database Underground that compared Oracle and SQL Server to the latter’s advantage. Sean follows up the original piece, with this second item on Oracle’s community. He writes, “. . . one area I think Oracle has it over Microsoft is in its downloads. When I go to Oracle to download anything, all the downloads are clearly marked on a single page. Microsoft just isn’t like that. Sometimes even finding a service pack for SQL is like finding help for Oracle. . . . to those of you who said Oracle is easier to admin that SQL, you’re just crazy.”
Sean continues the compare-and-contrast in another post on Oracle vs. SQL, concluding, “I’m not knocking Oracle as a platform. I never have. But come on guys, let’s be honest. You’re not the most user-friendly kid on the block.”
Perhaps Eddie Awad could speak to that. He notes yet another Oracle social network in the works, writing, “In addition to Oracle Mix, Oracle Wiki and Oracle Community, OAUG will be launching yet another Oracle related social network called the Knowledge Factory.” And Yas comments, “Another social network? It is getting really hard to keep up.”
Of course, there’s good material to be found if you can. On Oracle Community, shoblock offers a couple scripts for checking procedure dependencies: “Ever want to drop an old procedure, but you didn’t know if it was being used by some other code, perhaps in another schema? Simply check ALL_DEPENDENCIES, right. Except I hate typing those long column names. And what if I want to traverse through the dependencies?”
On Oracle Musings, Dominic Delmolino pitches in to an ongoing discussion of the value of information, revolving around the proposition: data, not programs, is the only thing that matters ”applications are transient and have no value except to acquire, manipulate, and display data. data is the only thing with value.” Dominic says, “I agree . . . that data that lies about, unexploited by any application, is a pretty useless waste of storage[, that] the true value of data comes from an ability to use it through an application which allows one to analyze, manipulate and visualize information synthesized from the data soup.” The piece looks at the value of SQL itself in this light.
The ORACLE-BASE Blog links to their in-depth article, SQL Plan Management in Oracle Database 11g Release 1.
Jason Arneil has been testing legacy applications and Oracle 11g. “I’ve seen [Oracle client] 8.0.5 running against, 9.2, 10.1 and 10.2 all very happily. However something has changed with 11g, and I cannot connect an 8.0.5 client to an 11g instance.” The article shows the what and the why, and possibly a workaround in the comments.
Richard Foote cracks what he calls the height reduction 1/2-myth, beginning, “A common misconception with using a larger block tablespace specifically for indexes is that this will result in a reduction in the height of indexes and hence ‘flatten’ index structures. However, this is only partly true, ” and ending, “In many databases, especially when the index block size is just doubled, it’s actually quite surprising just how unlikely it is for an index to actually decrease in height.” The middle part is the best.
Much of the talk in the PostgreSQL scene this week was about EnterpriseDB and IBM. On rand($thoughts);, Savio Rodrigues reports the news: IBM invests in EnterpriseDB.
Curt Monash delivers the Postgres Plus story: “EnterpriseDB is making a series of moves and announcements. Highlights include: renaming/repositioning the product as ‘Postgres Plus.’ The free product is now Postgres Plus, while the version you pay EnterpriseDB for is now Postgres Plus Advanced Server.”
Robert Treat’s analysis on xzillablog is, the enemy of my enemy is my friend, with the players being EnterpriseDB, IBM, MySQL, and Sun Microsystems.
In the Database Soup, Josh Berkus’s view on Postgres Plus and GridSQL: “For the last 5 years, I’ve been harassing all of the commercial PostgreSQL vendors — including the one I work for — to put together a serious PostgreSQL ‘distribution’, with a set of tools, installers and GUIs appropriate to the new user and competitive with the massive packages delivered by Microsoft and Oracle. . . . [It’s] time for us to get gussied up and go Red Hat or Ubuntu to compete with the big proprietary software. EnterpriseDB, if their press is to be believed, has finally done that. It’s about time someone did, and I’m glad they’re leading the way.”
Here’s a useful technical post from Sergey Petrunia on correlated semi-join subqueries and PostgreSQL
Some MySQL blogs now. Giuseppe Maxia, the Data Charmer, wants to know what kind of hardware you run MySQL on. Writes Giuseppe, “MySQL has a reputation of being a lightweight database. This definition can be intended either as having limited features or as requiring little hardware. . . . As for expensive hardware, my own experience is mixed. Yahoo! and Google have shown the world that using an array of inexpensive boxes in replication is often the most sensible way of scaling. However, there are many users who don’t feel comfortable with replication, and therefore buy more iron when their current box has reached its limits. . . . Given the choice, what is your favorite hardware for MySQL?” There is a poll in which you can cast your vote.
On the MooCow Productions Blog, a post in defense of surrogate keys. Tim Soderstrom writes, “Recently, I noticed a post on Planet MySQL expressing distaste for surrogate keys. Ever since reading it the topic has been bugging me so I thought I would finally break out a defense for surrogate keys. Really, it comes down to the right tool for the right job. Like The Force, they can be used for both good and evil. The fact is that a surrogate key can be wildly useful and efficient if used in the right context.” Tim builds his defense. From the jury box, Sheeri says, “If a surrogate key was a spice, it’d be salt. If it was a condiment, it’d be ketchup. It’s not that salt or ketchup aren’t good flavor enhancers, it’s that some people put them on EVERYTHING and it’s not the right answer for the problem at hand.”
And on the MySQL Performance Blog, Peter Zaitsev offers MySQL file system fragmentation benchmarks. Peter ran tests on several engines, and the comments shed a little more light on the facts.
Arjen Lentz offers his thoughts on Matt Asay’s post on open source’s “superficial impact” on the database market, in an item called The disruptor and the incumbent. “The assessment is correct (it’s based on facts), but the conclusions are simply irrelevant since they presume that there is a single correct objective or market – in the case of databases, ‘enterprise’ or ‘mission critical’ deployments. But just look at the situation: the definition of both is vague and continually shifting. Some companies don’t regard their website as mission critical, even though their online presence is critical to their survival. Go figure.”
So much my-dbms-versus-your-dbms stuff. Guy Bowerman weighs in on the database popularity contest, writing: “I love meaningless internet popularity contests where you never really know whether the winning object of affection has the popular vote or the most orchestrated get-the-votes-out campaign. This year IBM Informix Dynamic Server 11 (IDS 11) is in the finals of the CI Product of the Year: Databases, along with Oracle Database 11g, Microsoft SQL Server, MaxDB by MySQL and CodeGear’s InterBase 2007 Server Edition. So here’s your chance to vote. . . .”
It gets a little silly sometimes. How silly? How’bout taxi cab taunting and doctored photographs, as reported by Sean McCown. Funny stuff.
That’s all for now. See you in a week’s time!