Friday, October 31, 2008

11.50.xC3: another pack of features...

IBM has just made available IDS 11.50.xC3. Traditionally, fix packs include bug fixes and eventually some minor improvements. I believe the concept of "minor" may have to be revised when it comes to IDS fix packs. Let's see what the good fellows of R&D decided to put into this release:

  • MAX OS X improvements:
    • Silent installation
    • Ability to create Data Source Names (DSN)
  • Enterprise Replication improvements:
    • Dynamically change Enterprise Replication (ER) configuration parameters in the Onconfig file
      We can now do this using commands like cdr add onconfig, cdr change onconfig, and cdr remove onconfig
    • Improved Consistency Reporting after Enterprise Replication Repair Operations
      The repair jobs have an option that forces the retrying if inconsistencies are found at the end of the jobs
    • Administering and Monitoring Enterprise Replication with SQL Admin API
      The SQL admin now allows us to run cdr commands through SQL
  • Ability to change ONCONFIG parameters with SQL Admin API
    SET ONCONFIG for session change and SET ONCONFIG PERMANENT for Onconfig and session change (session here means the current instance configuration and not client session)
  • Three new dynamically changeable ONCONFIG parameters:
  • Improved SQL Tracing with the SQL Admin API
    From now on we can define which databases, sessions and users will be traced. This gives higher control on the SQL tracing facility. We can also suspend and resume the SQL tracing (without loosing previously set specifications)
  • Changing the Size definition of the First Extent of a Table
    Now we can establish a new definition for the first extent size of a table. Note that this will not change it's physical layout. It will be used if you extract the table schema, rebuild the table (by specifying a new fragmentation clause) or if you truncate the table with the reuse clause
  • Implementation of SAVEPOINTs
    SAVEPOINTS are an SQL standard that allows the programmer to establish some points after the begin of a transaction to where he can ROLLBACK. Imagine that you're running a batch file, processing several customers in a FOREACH cycle. You may want to do all the customers as a unit, so you would COMMIT at the end. But also consider you want to do several DML operations for each customer and that you accept that you can skip a customer, but not one or more operations within a customer. To put it simpler, if one operation on a customer fails you want to skip it, but continue with the transaction. With this feature it's possible. You just have to set a SAVEPOINT after each customer. If something goes wrong in the next customer, you can ROLLBACK WORK TO SAVEPOINT and continue with the next customer (maintaining your main transaction open).
  • Basic Text Search DataBlade Module Supports High-Availability Clusters
    In previous versions, the indexes created by the BTS needed to be created in external dbapces. This would mean that you couldn't use BTS searches on secondary nodes unless you used some (officially unsupported) external way of synchronizing these external spaces within the cluster hosts. Now, you can create these indexes in smart blob spaces, which can be logged, and thus can be replicated to all the cluster nodes
  • Querying XML Attributes with the Basic Text DataBlade Module
    BTS now supports another index creation parameter that specifies that the XML attributes (on XML tags) should be indexed (and thus become available for searches)
  • Setting the Frequency of Error Checking for Smart Large Object Transmission
    A new environment variable called IFX_LOB_XFERSIZE was introduced to specify the interval (in bytes) that will trigger error checking for SLOBs. After this amount of bytes the data will be checked for transmission error (between client and the database server). If it finds any errors it aborts with an error. Without this variable the checking is made only after the end of the transmission
A final note to the IDS Information Center. It now contains three new features:
  • An RSS feed with updates to the Information Center
  • Search engine add-ons for browsers, so that you can directly search on IBM Informix Information Center for your specific version
  • An extended section with links for community sites. This includes forums, general sites, wikis and blogs about Informix. The Informix-Techonology blog is one of the additions to this list. I face this as an honor and a great responsibility. It's also a great incentive to keep the work that I started more than one year ago.
I keep repeating myself about this idea: The IBM Informix team (management, architects, developers etc.) is doing a terrific job improving IDS while maintaining the ease of administration, reliability and performance we all know and love! With new features come some complexity increase, but that doesn't mean the product has to be slower, harder to manage or more unstable.
I look forward to the next release/fix pack which I bet will bring some more great improvements!

Wednesday, October 29, 2008

Encryption: All your (data) base are NOT belong to us!

If you were growing up during the personal computer boom you may notice that the title of this post is a joke with another joke. For details on that check the Wikipedia or the flash animation.
This is just a pretext to introduce a very serious subject. You probably know that Informix IDS can encrypt column data, but you may also know that it has some disadvantages (you cannot index encrypted data, you may have to change the applications etc.). There's nothing wrong with column encryption, but as with most of the features is has advantages and disadvantages.

Another option for encryption is to encrypt the files containing the data (chunks in IDS language). This brings up several questions:

  • Performance
    If you're doing data encryption you're consuming CPU cycles
  • Ease of use
    Is it transparent to the database? And to the application?
  • Privilege users access
    Can root work around the encryption? Usually, being root is like being God...
IBM has a solution for these kind of transparent low level encryption. It's a partnership with Vormetric which resulted in a product called IBM Database Encryption Expert. It tries to solve all the questions above. It's fast and light, meaning you won't have too much performance impact, it's easy to use and configure, through an administrative console, and you can stop the root user from accessing your encrypted data files.

IBM has just released fixpack 1.1.3 of this product, which is the first version to support IDS (versions 11.10FC2 and 11.50). You can get more info about it at the product page ( ).
Up to now, it was certified for use with DB2, so it's probable that you find several references of the two products together. One of them which I find interesting, is a video showing it working with DB2, by Belal Tassi, a colleague from IBM Information Management:

As you can see from the product page, the fixpack Readme and the video above, it's transparent for the database, for the application and it can be setup to be safe even from root. It's also able to encrypt backups (it would be a breach of security to have your data encrypted and keep your backups without encryption). Hint: Version 11.10 introduced backup and restore filters...
The platforms supported at the moment are:
  • AIX (64-bit):
  • Linux (64-bit): Red Hat Advanced Server 4.0 Update 4
  • Solaris 9
  • Solaris 10 update 2
  • Windows

Supported file systems

  • Windows: NTFS
  • Solaris: UFS, NFS, VxFS
  • RedHat: EXT3, NFS, VxFS
This product can be a solution for some compliance requirements, specially for companies who have to comply with Payment Card Industry Data Security Standards (PCI DSS), or simply for paranoid DBAs

Monday, October 27, 2008

Too many subjects...

Yes... Fortunately it's true: I can't keep up with everything that's been hapening....
In this post I'll mix a lot of topics. They've been covered elsewhere, but for the record, a few comments:

  • Carlton Doe, a well know name within the Informix community, has published a new book. It's called Administering Informix Dynamic Server: Building the Foundation. You can find it at Amazon. I don't have a copy yet, but I look forward to check on the contents. Given the author's history on Informix and previous work this book should be a safe investment
  • IBM has made available an Informix Virtual Appliance. This is essentially a virtual machine pre-configured with a lot of Informix software (IDS of course, Client SDK, JDBC driver, Spatial Datablade, Open Admin Tool, Data Server Drivers - JDBC/ODBC, and finally Data Studio). All running on top of SUSE Linux Enterprise Server 10 SP2. All you need to run it is a copy of one off the VMWare players. You can get a complete explanation and download instructions on Guy Bowerman's site
  • IBM started another initiative. It's called the Data Champion Program. The idea is to publicly recognize individuals who have made strong contributions to the Data Management community. The program website is located at Today, Mr. Arvind Krishan announced at IBM Information On Demand 2008 conference (IOD) the inaugural champions.
    I'd like to congratulate all of them, with just one special note to Eric Herber for his wonderful job at Informix Zone (
    You can nominate other champions on the program's website above until January 31, 2009.
  • Speaking about IOD, the IBM Information On Demand 2008 conference, it started yesterday and you can expect some important news. For now, on the light side, there are a lot of Informix tattooed people there... Take a look at this photo for example, courtesy of Jean George Perrin, showing a familiar face for the Portuguese Informix community, with a "quality stamp on the forehead" :)
  • One of the IOD news is about SolidDB. As you probably already know, Solid was acquired by IBM, and produces a product called SolidDB that can act as a cache for DB2 and Informix. SolidDB is an in-memory database, meaning it can speed things quite a bit. In this press release, IBM announces that it now supports RDBMS from other vendors. Specifically, SolidDB cache now supports DB2, Informix, Oracle, Sybase and in a near future SQL Server
  • And finally, an announcement from Exact, about plans to support for Informix in two products: iReplay (capture and replay SQL), and iWatch (SQL monitoring). You can check the press release here:
That's all for now. This post is just to keep you up to date. I'd love to return to long, boring, technical articles, but time is short... Hopefully soon...

Friday, October 17, 2008

Informix at VendorRate

If you follow some Informix blogs, groups or sites, you probably already know that Informix earned Top Customer Satisfaction Score on VendorRate survey in Q3. If you're into press releases you can learn more at Business Wire. If you prefer to check the report directly you can find it here.

Informix received the maximum score achieved by any vendor (91) and was rated by more than 120 organizations accordingly to the study. "O"ther RDBMS vendor got 81. The rating covers several aspects (Recommend, Integrity, Communication, Timeliness, Usability, Reliability, Customer Service, Budget, Expertise and Effectiveness). Informix got 9 in all of them except for Reliability (10) and Budget (8).

This rating reflects customer opinions. They are not marketing, although they're probably be used for marketing... I can't say I'm surprised. We all know that Informix customers are loyal, and they're loyal simply because they know they're using a solid product.

Obviously, we tend to compare "our" goods with the competition. And to be honest, we can't say the competition looks bad. As expected, a vendor who has the largest market share is well rated in general... But being a curious fellow I decided to dig a bit deeper... And if you follow me you can also check interesting points:

  • In the small companies category Informix is even better positioned. The direct competitor gets a lower rating than in the general rating. It doesn't even show on the table.
    This is not surprising considering the ease of administration and low total cost of ownership (TCO) of Informix
  • Informix continues to achieve top score in the medium size companies. In this category, the second RDBMS place doesn't belong to the biggest competitor
  • In the large enterprise category Informix still scores top ratings. The biggest competitor is in a close second place
This can lead to the conclusion that Informix is terribly good in all kinds of organizations. it simply does it's job, and you can count on top class vendor support.

Then, and just for fun, I decided to dig even deeper... I noticed that VendorRate was at the biggest competitor conference and issued a report based on data collected there.
Looking at this report we can also find very interesting (if not amusing) points:
  • Informix was rated exceptional by the attendees (page 5)
  • Informix was rated 2 points (93) above the quarterly VendorRate ratings (91) (page 10)
So, if people using our competition loves us, why shouldn't we?!

Wednesday, October 08, 2008

IIUG 2008 survey: again, time to speak up!

International Informix Users Group ( IIUG ) has put up another user survey where you can classify several proposed IDS new features. You can also name other features not specified in the survey.
Please note, that the fact that the presented features are available for voting doesn't necessarily mean they're being implemented. But your feedback is very important and IIUG will take it to R&D.
Many of the most voted features of last survey were in fact implemented in IDS 11.10 and IDS 11.50. So now it's time to make your voice be heard, and request that those annoying things be fixed. Some of my favorites that are not in the survey are:

  • Limit locks per session
    This is really an annoyance... Give the DBA more control please!
  • Clarification on support for Encryption Expert (IBM and Vormetric)
  • Compression
  • Configurable system page size
    This could be a step for the next one
  • Portable dbspaces (at least for same-indian and same page size system)
  • HDR between heterogeneous systems (same as previous)
  • Fine grained auditing
    Some control per object. Probably a big change and probably big performance impact... Maybe a switch per user/mask to keep the lighter current implementation...
  • End of user in OS limitation (default user for procedures/query plan?)
  • Storage provisioning
  • Some DW features available on XPS (bitmap indexes, star index...?)
Well, looking at the list again, I'd say that "asking is easy!..."
You can find the survey here:
Don't miss the opportunity to influence IDS future!