(Continued from Part 1)

 

In Part 1 I provided the background for my review this week of Edward Hurley's report on and Simon Williams's Lazy Software and so-called "Associative Model of Data" (AMD). Has Hurley educated himself on the subject, as I suggested? Or, at the very least, has he found a way to assess critically the vendor's claims? Let's see.

 

In Is It Time to Get Lazy? Hurley writes:

 

"UK-based Lazy Software has an industrious task in mind: Challenge the relational database. With less than 50 employees and about 45 customers, Lazy Software is trying to challenge the relational database supremacy of companies such as Oracle. Lazy' s Sentences database is based on the "associative model of data," the brainchild of Simon Williams, Lazy Software's CEO and co-founder. Williams believes the associative model makes developing programs much easier than relational databases."

 

When Hurley says 'database' he means, of course 'DBMS', which is not the same thing. And to the extent that Oracle has any supremacy, it is certainly not a relational supremacy: there are no truly relational DBMSs available, only SQL ones, which is hardly the same thing. Third, Williams does not challenge just DBMS products, but the relational model (RM) itself (see More On Williams's "AMD") that, given that RM is nothing but the application of logic to database management, is quite a tall order.

 

Before staking any claims of easier application development, it is incumbent on the author of any new data model purported to challenge RM to specify the theoretical foundation on which the new model rests. RM is predicate logic and set theory applied to database management, and it is that combination which guarantees correctness of databases and of results obtained from them, as well as providing analyzability of intent and thus optimizability for database operations. The only way to improve on that is to come up with a better theoretical foundation. In Part 1 I said that this is precisely the challenge I put to Williams: to specify formally and precisely what has he substituted for logic and mathematics that is better. Such a substitute would be a huge revolutionary development well beyond the scope of databases, which would warrant much more than just a short trade article. Williams has not responded to my challenge to date and I don't hold my breath.

 

'The company's name isn't a joke but plays off the company's mission. 'I had a mathematics teacher once tell me that all the great mathematicians were lazy. They looked for the simplest, easiest solution to a problem,' Williams said. 'Programmers should try to do the same thing.'"

 

So the problem is worse than I thought. Here we have Williams, who has learned mathematics at some point and who should certainly understand that RM is the only database approach based soundly on mathematics, yet he goes around claiming that he has managed to invent a better approach, which obviously does not have such a foundation. And for somebody who challenges RM, he ought to know that it is the simplest solution to the database problem – that is why it was invented in the first place.

 

"Lazy Software proclaims Sentences allows applications to be written much faster. "I found the issue of software reuse wasn't about better programming, but about the structure of the database," Williams said. Relational databases require applications to be rewritten when the database changes and vice versa. Sentences breaks that dependence chain, William said. Applications are independent of the Sentences database."

 

I do not know who crafted this paragraph, Williams, or Hurley, but whoever did, it is either confused, or uses poor terminology, or both; it betrays poor understanding of the fundamentals that characterizes most detractors of the RM:

 

·   The very concept of a database is predicated on data independence (DI) – the independence of applications from database functions. Any DBMS that fails to support any aspect of DI is not a DBMS with respect to that aspect, regardless of the data model underlying it.

 

·   One of the main advantages (and reason for being) of the RM is DI.

 

I venture to guess that Williams refers to current SQL DBMSs, which provide poor support of data independence. But that is not because they are relational, but because they are not relational enough! The solution is true RDBMSs (TRDBMS).

 

"Atlanta-based application service provider M2 Technologies is using Sentences to store the data for the Web sites and solutions it develops and hosts for chambers of commerce and other membership organizations. "For us, it's a matter of granularity. Sentences scales very well for large amounts of small bits of data," said Thomas P. Lennon Jr. M2's executive vice president of business development and strategic planning. "You also don't have to be a programming genius to get a huge database up and working." While the company is aiming its product towards relational database users, it realizes its place. A database like Sentences is to a relational database like a microwave oven is to a conventional oven, Williams said. "Each has its role, side-by-side. You wouldn't cook a TV dinner in an oven much as you wouldn't cook a 25-pound turkey in a microwave," he said. The newest version of Sentences, version 2, runs on Windows, Solaris, Linux and AIX platforms. It's also been certified IBM Server Proven with WebSphere on IBM's eServer iSeries, pSeries and xSeries. "

 

I have no idea what "granularity" and "scaling" means here; and how large are "large amounts" and how small are "small data"? What is clear is that this has absolutely nothing to do with the data model and everything to do with implementation details. As to the rest of the paragraph, I would suggest that those who understand ovens should work in that field and stay away from databases.

 

The main point here, though, is that this is the kind of material to be expected from a trade press completely ignorant of the subject matter that it covers. I dare anybody to find anything substantive in this article on the basis of which to assess the nature and/or value of the product being promoted. And given the state of knowledge in the field, how many would discern that the claims made for it are sheer nonsense?

 

 

Posted 08/18/02

 

 

 

[ABOUT] [QUOTES] [LINKS]