> The Internet is great for IMS, to put it simply. I should also point out > that IMS is ideally suited to the world of XML which, if one looks around > for half a second, is on the rise. XML is hierarchical, not relational, > and if you understand XML principles you basically understand IMS DB > principles. The world needs both forms of data storage. It's frankly > silly to claim otherwise, with all due respect to Gartner.
"Needs" is a stronger statement than I would make. There are trade-offs in both approaches, but DB2 didn't get to its position of dominance by being a lesser choice. It's better in most circumstances and "good enough" in the others. You can "get by" with either approach, but the relational one has proven the better over the last decade or two. > Relational > databases are fantastic and help solve a broad range of business problems, > but they aren't everything. The hierarchical data model is sometimes the fastest way to access data where the application fits a "get unique/get next" model but even that is not a certainty. The relational version of that is a select with a cursor and DB2 has sophisticated indexing and buffering that usually result in equivalent or better performance. As a database IMS DB is a bit user-vicious for developers and customers alike, but it does what it does extremely well. On the other hand DB2 version 9 has some really cool stuff specifically for dealing with XML. And DB2 runs high transaction rates and it is a lot easier to deal with both on the application side and the operational side. If you were going to bet your paycheck on a database, would you pick IMS over DB2? > IMS is both a transaction manager and a database. There was no mention in > the article of IMS TM. Based on nothing more than customer experience and my own past life, I would guess a significant fraction of IMS TM applications actually use DB2 as the database, just as it is with WAS and CICS. Of course, you can access just about anything from anything these days so that's not such a big deal. > Lastly, IMS is extremely efficient. That's very important for high volume > applications, and the world requires more high volume applications each > year, not less. It depends on how you measure efficiency. For processing efficiency, as noted above, "It depends". If an application runs 50% faster, but care and feeding of the platform takes more people time, then more than likely that application is less economically efficient, all else considered. If the application has been around for years, then the platform isn't usually an issue whereas it probably would be if you were planning to build a new application. And that's where IMS is right now. It is great for "legacy" (I hate that word) stuff and it is preserving ROI by evolving and adapting to fit new ways of doing things. Realistically it's never going to be the belle of the ball anymore. CC ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
