Relational Data Model: Back to the Roots
Relational Data Model: Back to the Roots
The R3DM/S3DM framework we propose to be adopted in DB management systems can be considered in many ways an extension to Codd’s relational model. I am aware that this is a big claim.
Join the DZone community and get the full member experience.Join For Free
Download the Altoros NoSQL Performance Benchmark 2018. Compare top NoSQL solutions – Couchbase Server v5.5, MongoDB v3.6, and DataStax Enterprise v6 (Cassandra).
Back to the Roots
Recently, I realized that the associative/semiotic/hypergraph (R3DM/S3DM) technology framework we propose to be adopted in database management systems can be considered in many ways an extension to Codd’srelational model. I am aware that this is a big claim and certainly this is not the place or the time to lay down my arguments, but this is how it occurred to me.
I have partially implemented TRIADB technology twice on top of two different data stores and I noticed that the
GET operations we defined were closely related to Codd’s relational algebra operations, while datasets, i.e. domains, and a user-defined type system match the sets defined in mathematics and relational theory. Coincidentally, Codd’s relational logic goes back to Aristotle and the cornerstone of our technology, which is the computational semiotic triangle. I will briefly mention that one basic difference is that both the heading set and the body tuples of the relation — in fact, everything — are transformed and uniformly represented with numerical key references. Therefore, it can also be called a reference database management system (RDBMS). All these are simply good indications. I believe we are on the right track.
The truth is, and I will quote Chris Date here, that:
If you are proposing to replace technology A with technology B, you must understand technology A and there must be some problem that technology A does not solve and that technology B does solve.
And the best person I have found to teach me an in-depth understanding of relational database technology is Chris Date himself. The following video is a clip from an excellent illuminating workshop that explains Codd’s Relational Theory for computer professionals, but most importantly, he shows what a real relational product would be like and how and why it would be so much better than what’s currently available.
Relational Model vs. Other Data Models
That said, allow me to have my doubts about whether many of the proponents of other database technologies, including those in SQL databases and those in NoSQL databases, have understood what the differences really are with respect to relational models and at what abstraction level they occur. Again, this is not the place or time to elaborate on this. Instead, I am inviting you to ponder the architectural design of modern database management systems.
You see, in practice, it is too difficult to make a very clean separation between the physical, logical, and conceptual levels of information. From an engineer’s point-of-view, it is hard to separate theoretical from practical purposes. Moreover, many of these NoSQL DBMSs that are in fashion are suited to solve a particular type of problem. This is why you often hear that big corporations and large companies have many different kinds of DBMSs at the back-end — not to mention that there is a trend to market many DBMSs as multi-model database systems. And that made me also realize that there has to be a distinction between those problems a problem that solves at the physical level, i.e. like partitioning and availability, and at the logical-conceptual level, i.e. integrity and data modeling. Therefore, I foresee that in the future, systems will have to use a combination of these two levels that somehow will have to be tuned and made to work harmonically independent of each other.
This is our perspective toward the architectural design of modern database management systems that fully justifies our choice of marketing TRIADB as a middleware. We are focusing on providing an efficient and effective solution at the logical and conceptual level using an existing implementation of the database physical layer. Relational modeling theory applies here, too. From what I understand, it was the implementation details at the physical level and perhaps other naive simplifications that made many depart from the original relational model. So, it’s time to go back to the roots and make some real progress.
In case you, as a reader, have the same feelings and see some truth in my writing, I would be more than happy to discuss with you the progress we are making with TRIADB and associative, semiotic, hypergraph technology and definitely exchange ideas and share some common thoughts on these database topics. Stay tuned.
The fact that a DATA ALGEBRA never existed until a handful of years ago will likely surprise you. If you are not a software engineer, you probably assume that such an algebra was developed many decades ago and is regularly employed by software engineers to resolve tricky engineering problems they encounter. - Garry Sherman, Robin Bloor
Published at DZone with permission of Athanassios I. Hatzis, PhD , DZone MVB. See the original article here.
Opinions expressed by DZone contributors are their own.