From: Neva Cherniavsky (nchernia@cs.washington.edu)
Date: Mon Apr 26 2004 - 08:29:10 PDT
This paper describes the state of the database world in 1986, what changed
in intervening ten years, and where the research will go in the future.
The paper was written in 1996, and so provides an interesting snapshot of
the past.
The authors identify four active areas of database research going on in
1986. These are extended relational database systems, persistent
programming languages, object-oriented database systems, and database
system toolkits/components. Of these four, the first was the commercial
winner in 1996; they included objects without sacrificing the underlying
relational model and query language, and were thus simple and natural
extensions of existing technology. Persistent programming languages,
while still an active academic research area, failed to make an impact on
commercial applications. Database system toolkits, a research interest of
the authors, were both too complex for the applications programmer and not
complex enough for the database implementor.
Perhaps the most surprising loser (surprising to the researchers in 1986)
was object oriented databases. One big problem was that researchers
differed in nearly every single area in which people could differ:
programming interfaces, implementation, and query support. Another
problem was that OODB didn't manage to support everything relational
databases support, such as views, and were not as robust and
fault-tolerant as their counterparts. Lastly, OODB systems were tightly
bound to their implementation language, making application programming
less flexible.
The authors think the future is in the complete ascension of
object-relational databases over OODB, and that OODB will remain a niche
area. The latter point seems to be true, though I don't know enough of
the current state of the research to confirm or deny the former. They
also recommend throwing out SQL and starting over, which I doubt has
happened. I liked their thoughts on middleware that can help access
different types of legacy databases. Meanwhile, the authors completely
missed the impact that the internet would have on databases, particularly
in the transmission of data. This seems to be an active area of research
currently, with XML databases garnering a lot of interest. They can be
excused for not seeing the future. In some ways I wish the authors had
taken a less informal tone, but this sort of survey is both interesting to
future generations and helpful to the current one.
This archive was generated by hypermail 2.1.6 : Mon Apr 26 2004 - 08:29:11 PDT