Oracle + Open Source
two worlds inexorably colliding

As many of you may have already heard, Sun officially announced their purchase of  MySQL today.  After the last few years with Oracle pushing Linux and commodity hardware, Sun has certainly taken the hit.  I guess this is their turn to hit back.

With MySQL 6.0 out, increasingly we find the full compliment of sophisticated database features in MySQL.  But a lot of the devil is in the details.  Where Oracle has had problems with the sheer size of the codebase, and addressing security vulnerabilities, and other bugs in a timely manner, MySQL has the problem of a mature codebase.  Some of these features are newly available, and if my experiences with replication are any indication, often have hidden gotchas and “features” which are not emphasized in the literature.

The next question on my mind is, how does Oracle’s purchase and now ownership of Innobase impact the above purchase.  It means a direct competitor owns a core component which provides transactional support to your database.  A very good question.

Time will tell, so stay tuned.

The folks over at zmanda who specialize in MySQL backup are doing a webinar on December 13th, this Thursday.  They specifically mention it as relevant for Oracle DBAs or anyone moving to MySQL.  Looks to be an interesting event.

Incredible, but true.  It seems that MySQL AB will be exhibiting next week at Oracle Open World.  This of course isn’t the first time a competitor would advertise or exhibit on it’s rival’s home turf.  Still it certainly signals a changing landscape, and heats up the battle for market share.

Here’s a longer list of exhibitors at the conference.  I don’t see Enterprise DB there, but anything’s possible.  You will see RedHat as well as Suse, now owned by Novell, represented there as well.  Also if you make it to the conference, be sure to visit the Oracle pavilion section, where there are sure to be smaller booths for the Open Source Group, as well as Oracle Unbreakable Linux Support program.

The fifth and last abstract submitted for the O’Reilly MySQL Conference in April 2008.

As an independent consultant, there are quite a few trouble spots I see repeatedly. I’ll discuss five of them, and how to avoid them in your own infrastructure.

As an independent consultant for twelve years, I’ve encountered a lot of interesting and challenging projects. I’ll discuss five different cases, and what lessons I took away from each.

1. Intro
2. The Right Hardware
3. Importance of Good Testing
4. Patchwork or Good Design
5. Don’t Mix Opposites
6. Use The Technology
7. Conclusion

The fourth in a series of five abstracts for the O’Reilly MySQL Conference in April 2008.

Learn how to audit your systems, and run through the right checklists so you can sleep better at night knowing your systems are more secure.

Security is on everyone’s radar these days. You may be wondering yourself whether your database systems are really as secure as they should be. We’ll discuss some of the latest vulnerabilities, and what you can do to protect your systems.

1. Introduction
2. Authentication
3. SQL Injection
4. OS Security
5. Network Security
6. Conclusions

This is the third in a series of five abstracts submitted to the O’Reilly MySQL Conference in April 2008.

MySQL has a great facility for creating a read-only failover database. We’ll show you how to setup, start, failover, and monitor it.

Setting up MySQL to have a master + slave failover capability might be intimidating, but it needn’t be.

1. Intro
2. Anatomy of MySQL Replication
3. Initial Master copy
4. Setup + starting the slave
5. Failover from Master
6. Adding another slave
7. Monitoring your slave db
8. Conclusions

The second in a series of five abstracts for the O’Reilly MySQL Conference in April 2008.

Inevitably hackers are trying to get at your data, so you mine as well know what they can and can’t do. What better way to discover where you’re vulnerable than hacking your own systems.

Operating Systems have bugs, Database Software has bugs, and so does your application, probably. A better question is how hackable are you? We’ll look at some of the nefarious ways intruders can get in, so you’ll better know how secure your systems really are.

1. Intro
2. OS level
3. Database level
4. Application level
5. Conclusions

I’ve just put together my abstracts for O’Reilly’s MySQL Conference in April 2008. Some of them might sound familiar…

Learn to watch your database like a fitness diet. Trim down the SQL queries, use the right hardware, and monitor the right metrics to keep it running fast.

There are healthy databases and their are unhealthy ones. We’ll take a look at what you feed your database, and how to keep it fit with just the right diet of hardware, configuration, and SQL query tuning.

1. Introduction - Diet of a Champion Database
2. Disk, Memory, CPU - Body by Intel
3. Applications - Lean & Fit
4. SQL Queries - High Fiber, Low Fat
5. Conclusions

Typically a backup or dump of a MySQL server includes all of the databases available, using the -A or –all-databases options. But what of restoring, and recovering that dump?

One can simply go to the target machine, and delete everything in the data directory, right? Oops, you didn’t delete the initial MySQL database did you? How about the special “information_schema”, MySQL’s data dictionary? The other option of course is to use mysqladmin:

SQL> mysqladmin -f -u root -p drop mydatabase

But still I’ve had cases where I’ve dropped parts or all of these initial MySQL databases. So what to do if you do?

Luckily MySQL comes with a shell script to save you in just such cases. It’s called mysql_install_db and can typically be found in /usr/bin. For Oracle folks you can almost think of this like the catalog.sql which in turn runs the sql.bsq file. It is illustrative to take a look at this shell script, and see what’s contained in there. You’ll learn a lot about the bootstrapping process.

MySQL has documentation on the mysql_install_db script.

So when you go to building your backup scripts, and are putting all the pieces in place, be sure to make a note of this script, and remember where it is. If you are providing instructions for recovery for Unix Admins who may not know MySQL particularly well, be sure there is a note, or even better yet, a call to this script in your own restore scripts.

In Part II in this series, I talk about how these three databases compare in some particularly crucial areas.

For instance how do the optimizers of these different database engines behave, and why does that matter?  What type of indexes are available, particularly with respect to typical applications.  I then move on to datatypes available and which are missing.  You’ll find some surprises here.

Lastly the holy grail of any modern relational database, I discuss transactional support. Relevant concepts include ACID compliance, read-only versus insert and update activity, and so on.

Recently I wanted to setup a little MySQL sandbox where I could hack away at MySQL with reckless abandon.? A sandbox is different than a test environment, it’s usually one which is very breakable.? You want to be able to break things, or rather take them completely apart and put them back together.? It’s the only way to understand all of the moving parts.

So searching google, I happened upon Giuseppe Maxia’s Replication Playground. It basically installs into an unprivileged directory, one master, and three slaves.? You can then test out various scenarios. Read his blog entry.

It is trivial to install, however I encountered some issues with MySQL 5.0, which caused me some troubles.? I sent him my feedback, and comments, and it looks like he has rereleased it as the MySQL Sandbox. Good stuff.

If you’re interested in how these three databases measure up in terms of feature sets, take a look at part one in a two part series I wrote over at Database Journal.

I discuss stored procedures, views, materialized views or snapshots, triggers, and security. Stored procedures and functions are supported on all three databases, as are views and triggers. Although MySQL and Postgres aren’t there in terms of default snapshot support, there are ways to get that functionality in a somewhat roundabout way.

Security is always a tricky question, as all the bugs out there aren’t always publicized. It’s sort of a cat and mouse game. All three databases support user based authentication to login to the database, and various privilege levels to control access to objects and data. Oracle also supports FGA or fine grained access control for column level control.

That said I might tend to say that open-source products in general have better security, their source being an open book and all.