IBM and Informix

I have periodically written about Informix Dynamic Server (IDS) ever since IBM acquired it. Initially, IBM had the wrong messaging—all of its databases were marketed under the DB2 brand, which didn’t go down too well—but this has now changed with the company’s refocus on information management in general and Information on Demand in particular.

Then IBM came out with IDS 10 but, while this fulfilled the company’s promise to users to continue to develop IDS, it was hardly the most exciting release ever made. At the time I described it as “worthy”, which is about as close as you can get to damning something with faint praise.

Now, IBM has announced the latest release of IDS, code-named Cheetah, formally IDS 11. In contrast to IDS 10, which was primarily about advances and enhancements to technical features, the main emphasis in this release is on availability. In particular, it has introduced an “availability fabric” that allows you to designate one or more secondary servers, which may be either local or remote, as failover-ready. Moreover, if you have multiple servers installed then you can take any of these off-line as required. This means that you can maintain availability during both planned (for example, you can take off-line one server at a time in the case of rolling software upgrades) and unplanned outages.

Of course, high availability is no use without performance. If your system is available but takes 5 minutes to present you with the information that would normally take 2 seconds you might as well call it unavailable. Anyway, there are some significant performance enhancements in this release: one example IBM quoted to me was a function whose performance was improved by more than 5 times—though I expect this is optimal.

Some other major enhancements include support for label-based access control that has been ported across from the DB2 environment, a new web-based administrative tool that can be used against IDS instances, application development improvements, extended support for location-based services and a number of other features.

IBM sees IDS’ primary positioning as “the optimal data server for integrated relational OLTP environments“. That is, OLTP environments where the database is embedded in the application. IDS has always been well-known as a high performance database that requires very little in the way of administration and it is this latter point that makes it attractive to software developers and partners: what is sometimes called “fire and forget”. Of course, much of the partner market is in the SMB space and for companies in this area of the market features such as the availability fabric are perhaps less relevant. Nevertheless, for larger organisations and for those that need real-time capabilities (of which there are a number of IDS users) such functionality will be crucial.

When IBM first acquired IDS it intimated that in a decade or so the two code lines might merge. After two major releases of IDS this doesn’t look likely and it seems as if IDS will continue to have a significant future as a product in its own right. Indeed, IBM claims that sales of Informix are growing faster than the database market as a whole: and long may it be so.