Re: Recommended papers / books
Date: Sat, 10 Apr 2004 08:06:05 GMT
"Scottie Swenson" <swenson_at_heronetwork.com> wrote in message
> I am doing research into distributed database designs. I have run into a
> few scattered papers and even fewer books.
I can give my impressions, but they might be different from others in this group.
Sheth and Larson wrote a paper on federated database systems (Sheth, A. & Larson, J., (1990)) that I believe is regarded as an excellent formal treatment of federated database system architectures across the dimensions of heterogeneity, distribution, and autonomy. After Date, which is rather limited in the *Intro* text, this was my first exposure to a formal treatment of the subject.
Sheth, A. & Larson, J., (1990). Federated Database Systems for Managing
Distributed, Heterogeneous, and Autonomous Databases. ACM Computing
> I am considering getting "Principles of Distributed Database Systems" 2nd
> Edition, by M. Tamer Ozsu and Patrick Valduriez. But, the reviews for this
> books are weak at best. The reviewers mention that there are better
> references without actually pointing out what those may be.
Who were the reviewers? This was a textbook I used in a graduate course, and though I didn't agree with everything, I haven't seen another book or reference that comes close in terms of comprehensiveness concerning the myriad of issues, historical precedents, and architectural possibilities. For me, it also highlighted the vast complexities of object-oriented data management in a distributed system, something that I doubt the authors wanted to convey specifically. One thing that I found a little different was that federated architectures with high degrees of autonomy were placed well in the back of the text.
The bibliographic notes were very good also and would give you a map to a wide varieity of relevant research materials.
Some of the issues it covers are:
1) types or models of distributed architectures 2) topics on distributed relational systems, including distributed optimization, integrity and concurrency issues; placement of data and other design issues; fragmentation and redundancy, etc. It also gives a historical comparisons of Ingres and R* distributed algorithms and approaches.
3) Distributed object-oriented systems, touching on some of the same design issues as the relational counterparts.
It's somewhat formal and theoretical, but useful in a practical sense as well, especially in terms of fragmentation and redundancy with relational databases. However, it seems geared more towards systems designers. If you want a practical handbook relevant to applications in today's implementations, this might not be the best, since many extensions would have to be hand-coded on top of or wrapped on native DBMS functionality. It's more of an overview of a wide variety of approaches.
The database research community seems beholden with the idea of the mediator approach in distributed database management and data integration these days. Professor Gio Wiederhold from Stanford wrote one of the first, perhaps seminal, papers on this subject and research has followed, such as TSIMMIS and others.
Wiederhold, G. (1992). Mediators in the Architecture of Future Information
Systems. IEEE Computer.
Dan Received on Sat Apr 10 2004 - 10:06:05 CEST