Caché vs Oracle vs RDBMS - Please add your 2 cents!

Question

Hi All,
I have recently “discovered” the Caché DBMS and realised just how little information there is out there. Apart from a few forums and a few vendor websites (mostly InterSystems) I haven’t been able to find much documentation or FAQ’s or user guides/manuals about the MUMPS (and Caché) database. HOWEVER(!) I have found two articles (funnily enough published by the Caché vendor InterSystems) about how much faster Caché is compared to Oracle. So the reason I’m posting here to (hopefully) start a discussion and get some input from Oracle users out there who have user Caché.
There are many claims by Caché users about how ultra-fast it is, how easy it to write application code (since the DBMS is object oriented), how the databases run with next to no intervention by DBA's, etc, etc. So if, as they claim, it's such a fantastic product how come 99% of the time I draw a blank when I ask other people in the field about Caché? Absolutely no one in my organisation knows how to use the Caché DB that we use. The end users access the Application...and that's it! And if the Caché technology is genuinely fast and efficient shouldn't the big 3 (Oracle, MS, IBM) include Caché like technology in their products or at least introduce competing?
Don't get me wrong. This is not an attempt to say one DBMS is better than the other. I just want to see how many have heard of/used Caché and what the general consensus is about Caché. And what do you think about OO-DBMS, is it the future? There's a link below that says how OO-DBMS' are the best options for large-scare Internet systems (citing Google's BigTable and Amazon's SimpleDB), do you agree?
Some links:
http://www.computerworld.com/s/article/101591/Looking_Beyond_the_Big_Three
http://labsoftnews.typepad.com/lab_soft_news/2010/05/persistence-of-the-mumps-programming-language-in-healthcare.html
http://labsoftnews.typepad.com/lab_soft_news/2010/10/more-on-mumps.html
http://www.faqs.org/faqs/m-technology-faq/part1/
http://www.cs.uni.edu/~okane/source/ISR/isr.html#ex01
http://www.cachewiki.org/index.php/About
http://bytes.com/topic/oracle/answers/65655-comments-intersystems-cache-database-platform
http://dbaspot.com/forums/object-database-technologies/196842-help-problems-intersystems-cache.html
http://www.intersystems.com/cache/whitepapers/pdf/BestKeptSecret.pdf
http://www.outoftheslipstream.com/node/124
http://www.dbms2.com/2010/01/15/intersystems-cache-highlights/#more-1400
Edited by: feersum_endjinn on 15-Apr-2011 07:28
Edited by: feersum_endjinn on 15-Apr-2011 07:31

Answer 1

Oh, you mean [url http://isabel.la/?q=node/24]DSM-11? ;)
I think if you look around you will find Oracle is used very heavily in the health care field. While there is nothing wrong with software written for the ages, surely more should be, trying to keep stuff written that long ago creeping along is just way too expensive, it has too many implicit assumptions about how systems should be designed and the relative costs of the various trade-offs. Of course, some of the [url http://web.signonsandiego.com/news/2011/feb/20/wireless-device-would-give-doctors-access-real-tim/]newest stuff is being designed haphazardly, which has a great deal of risk. Add in modern legislative issues and you have a real mess.
Claims of 4GL or object-oriented development being faster are misleading. Two things wind up happening: compromises are made as technology develops, and maintenance becomes increasingly difficult because it is too easy to change things. In addition, design philosophies spring up that have strange ideas about how to deconstruct a large system into small component parts. As the others noted, we periodically see this new paradigm baloney, and it goes nowhere.
The speed of various paradigms can be better than a pure relational approach - see [url http://kevinclosson.wordpress.com/kevin-closson-index/exadata-posts/]Exadata, for one - but they inevitably need a very specific approach to very specific problems. It gets to the point where vendors require very specific hardware and software requirements to make it work - see [url http://kerryosborne.oracle-guy.com/2010/08/running-oracle-exadata-v2-on-dell-hardware/]Exadata, for one.
Large scale systems like google are BS for actual applications. What is google? It is an advertising company. Look at it's SEC filings, it says that in so many words. The data it presents does not have to be accurate or consistent, and it is not. It doesn't take much effort at all to find searches that don't work, or give different results. All those other google apps you hear so much about have not proven themselves to be successful, and most are failures in a business sense. Follow the money.
Amazon is a bit better example, if you happen to be an Amazon size company. Note that the actual business runs on Oracle, and was able to grow as it did by using Oracle. All that "eventually transactionally consistent" stuff has to end up somewhere. As for simpledb, what real use is being able to change your data model on the fly? You need to fix your modeling skills!
Larry Ellison is a very bright guy, and as technologies prove themselves they get brought in and others get brought in before they prove themselves. Some of them fail and are quietly ignored, some are simply too far ahead of their time. Everyone wants to be the next google, but most fail.