Follow @DBDebunk Follow @ThePostWest

Note: To demonstrate the correctness and stability due to a sound theoretical foundation relative to the industry's fad-driven "cookbook" practices, I am re-publishing as "Oldies But Goodies" material from the old DBDebunk.com (2000-06), so that you can judge for yourself how well my arguments hold up and whether the industry has progressed beyond the misconceptions those arguments were intended to dispel. I may revise, break into parts, and/or add comments and/or references.

Remember XML DBMS? At one point it was the fad of the day, similar to today's NoSQL or the old new "knowledge graph" -- "the future that you ignored at the peril of being left behind". As I predicted, it went the way of all fads (ODBMS, Associative DBMS, you name them) together with their "data models" that were nothing of the sort. My prediction was grounded in the same sound foundations I rely on today -- unlike the industry we are progressing it -- that fads lack and which were and still are dismissed, evidence be damned.

Here's a typical example (comments on republication in square brackets).

Follow @DBDebunk Follow @ThePostWest

 with David McGoveran 

(Title inspired by Richard Feynman)

In Part 1 we explained that all database relations are, mathematically, relations, but not all relations are database relations, which are in both 1NF and 5NF and we agreed with a statement in a LinkedIn discussion ending as follows: "Update anomalies are not as big of a problem as an algebra where relations aren't closed under join". Unfortunately, update anomalies, closure, and how relational operators were defined are all interrelated and represent an even "bigger problem". Update anomalies are not "bugs", let alone irrelevant, but actually a reflection of  that much bigger problem.

In this second part we delve into that problem.

Follow @DBDebunk Follow @ThePostWest

Note: To demonstrate the correctness and stability due to a sound theoretical foundation relative to the industry's fad-driven "cookbook" practices, I am re-publishing as "Oldies But Goodies" material from the old DBDebunk.com (2000-06), so that you can judge for yourself how well my arguments hold up and whether the industry has progressed beyond the misconceptions those arguments were intended to dispel. I may slightly revise, break into parts, and/or add comments and/or references.

This is an email exchange with a reader responding to my third book.
(Originally posted on 06/21/2001)

“I'm presently reading your book PRACTICAL ISSUES IN DATABASE MANAGEMENT and there are a couple of points that I find a little confusing. I'll start first by saying that I have no formal database oriented education, and I'm attempting to familiarize myself with some of the underlying theories and practices, so that I can further my personal education and career prospects (but aren't we all!). My questions may sound a little bit ignorant, but that would be because I am! (Please note ignorant, not stupid!) I'll quote you directly from the book for this (possibly I'm taking you out of context or missing something important)

Chapter 3, A Matter of Identity: Keys, pg. 75: "Databases represent assertions of fact - propositions - about entities of interest in the real world. The representation must be correct - only true propositions (facts) must be represented."

Now, correct me if I'm wrong with a basic assumption here, but isn't a database simply a model of a "real world" data collection? I would've thought that the intention of a database would be to model real life effectively (and accurately) enough to provide useful data for interpretation. Now obviously this is not an easy process with complex data types, but would it even be possible to have a 100% true proposition with only atomic data types? (i.e. can a simplified model contain only facts?) In my understanding of modeling, any model that fits real life closely enough to be a good statistical representation is a usable model. e.g. Newton's Laws are accurate enough when applied on a local scale, but we need to use Einstein's model of space-time across larger scales. Wouldn't recording only "facts" (which I would presume you mean to be statements that are provable in the objective sense i.e. no interpretation, only investigation or calculation) possibly eliminate the utility of some aspects of the database? Or do we account for the interpretative aspect in the metadata or in some other way?

Essentially, I can see what you're saying, but not necessarily how you've reached the conclusion. Admittedly in an ideal world we should be able to record only facts in a database, but this is not an ideal world. As an example, in surveys we see such questions as "Are you happy with this product?" followed by a rating system of 1-5, or 'completely unhappy to completely happy'. This is an artificial enforcement of a quantitative measure on a qualitative property. How do we account for the fact that this is interpreted data and not calculated or measured?

My questions may have little relevance to database theory in general, but the concept fascinates me!”

Follow @DBDebunk Follow @ThePostWest

 with David McGoveran 

(Title inspired by Richard Feynman)

“As currently defined, relational algebra produces anomalies when applied to non-5NF relations. Since an algebra cannot have anomalies, they should have raised a red flag that RA was not defined quite right, especially defining "relation" as a 1NF table and claiming algebraic closure because 1NF was preserved. Being restricted to tabular representation as the "language" for relationships is like being restricted to arithmetic when doing higher mathematics like differential calculus -- you need more expressive power, not less! Defining RA operations in terms of table manipulations aided initial learning and implementations by making data management look simple and VISUAL. Unfortunately, it was never grasped how much was missing, let alone how much more "intelligent" the RA and the RDBMS needed to be made to fix the problems. And I can see that those oversights were, in part, probably due to having to spend so much time correcting the ignorance in the industry."

--David McGoveran

I recently posted the following Fundamental Truth of the Week on LinkedIn, together with links to more detailed discussions of 1NF and 5NF (see References):

“According to conventional understanding of the RDM (such as it is) [and I don't mean SQL], a relation is in at least first normal form (1NF) -- it has only attributes drawn from simple domains (i.e., no "nested relations") -- the formal way of saying that a relation represents at the logical level an entity group from the conceptual level that has only individual entities -- no groups thereof -- as members. 1NF is required for decidability of the data sublanguage.

However, correctness, namely (1) system-guaranteed logical validity (i.e., query results follow provably from the database) and (2) by-design semantic consistency (of query results with the conceptual model) requires that relations are in both 1NF and fifth normal form (5NF). Formally, the only dependencies that hold in a 5NF relation are functional dependencies of non-key attributes on the PK -- for each PK value there is exactly one value of every corresponding non-key attribute value. This is the formal way of saying that a relation represents facts about a group of entities of a single type.

Therefore we now contend that database relations are BY DEFINITION in both 1NF and 5NF, otherwise all bets are off.”

It triggered a discussion that raised some fundamental issues for which an online exchange is too limiting. This post offers further clarifications, including comments by David McGoveran, on whose interpretation of the RDM (LOGIC FOR SERIOUS DATABASE FOLKS, forthcoming) I rely on. The portions of my interlocutor in the discussion are in quotes.

Follow @DBDebunk Follow @ThePostWest

“Relational Data Integrity is like every other integrity constraint that checks that the relationships created between data using foreign keys has a consistency. This can be done by using ON UPDATE, ON DELETE constraints on the table.”

--Quora.com

I recently quoted this as one of my To Laugh or Cry? items on LinkedIn, which initiated an exchange triggered by the following question:

“You have a better definition? What is it?”

In the exchange the asker's interpretation seemed to be "referential constraints are constraints like any other constraints, so there is no problem".  It is hard to recognize misconceptions without proper understanding of the RDM. We ignore that the above is not really a definition and focus on debunking.

Decades ago I wrote an article in DATABASE PROGRAMMING AND DESIGN carrying the double-meaning title Integrity Is Not Only Referential, in which I debunked Borland's claim that its Paradox file manager supported referential integrity (at the time no PC product did). As one component of the RDM, database integrity is, of course, a DBMS function, but Pradox relegated it to applications. Then, as now, one of the most common and entrenched misconceptions was that relational comes from "relationships between tables" and so relational integrity amounts to referential integrity (RI). RI is, of course, but one of several components that comprise relational integrity -- it is necessary, but insufficient. While practitioners are familiar with referential and PK constraints, if asked what other constraints comprise relational integrity  very few know. Having enumerated them recently on LinkedIn, I asked this very question:

“... what other RELATIONAL constraints ARE there and what is their purpose? I recently posted a weekly truth and other items here that answer it.”

which went unanswered.

Data integrity is one of the three components of the RDM, together with data structure and manipulation. It consists of several categories of constraints which I detailed more than once, most recently in Understanding Relational Constraints, to which I referred the asker (can you give an example for each category?) Defining relational integrity means specifying all the constraint categories required by the RDM.

Consider now the above paragraph: it purports to define relational integrity, but it specifies functionality of referential integrity -- implying the old misconception I wrote about decades ago. The asker did not seem to comprehend the distinction:

“I can't see a problem here. Isn't it simply as follows? ... A *referential integrity constraint* ensures consistency between attributes of different entities - e.g. between primary and foreign keys of related entities (aka relational integrity). Isn't that what the definition says?"

Yes, it is the definition of referential integrity, but not of relational integrity -- there is more to the latter than the former. No matter in how many ways I tried to explain this, I was unable to convey it, because it's practically impossible in the absence of sufficient knowledge and understanding of the RDM.

 

 

 

 

Follow @DBDebunk Follow @ThePostWest

Note: Each "Test Your Foundation Knowledge" post presents one or more misconceptions about data fundamentals. To test your knowledge, first try to detect them, then proceed to read our debunking, reflecting the current understanding of the RDM, distinct from whatever has passed for it in the industry to date. If there isn't a match, you can review references -- reflecting the current understanding of the RDM, distinct from whatever has passed for it in the industry to date -- which explain and correct the misconceptions. You can acquire further knowledge by checking out our POSTS, BOOKS, PAPERS, LINKS (or, better, organize one of our on-site SEMINARS, which can be customized to specific needs).

“If you have shopping cart, you probably have some field "TOTAL" somewhere that stores the final amount due for the customer. It so happens that such a thing violates relational theory...”

“Having a "TOTAL" field in your "order" table *might* violate relational theory, but if you make it so that only a trigger can update it based on what's in your "order_item" table, then I think it's fine. You still get data integrity and that is what matters.”

“I still fail to see what you mean by the "calculated TOTALS field" (attribute, really) violates the Relational Model.”

“The result of having the field ... is what is called a DELETE ANOMALY.”

“Most denormalizing means adding columns to tables that provide values you would otherwise have to calculate as needed.”

“There are four practical problems with a fully normalized database, three of which I have listed before. I will list them all here for completeness:
* No calculated values. Calculated values are a fact of life for all applications, but a normalized database lacks them. The burden of providing calculated values must be taken up by somebody somehow. Denormalization is one approach to this, though there are others.”

--Database Programmer blog

“...I'm now working with IT to normalize part of the database to remove calculated fields...:
`lineitems`.`extended total` = `lineitems`.`units` * `biditems`.`price`.
`jobs`.`jobvalue` = the sum of related `lineitems`.`extended total` records
`orders`.`ordervalue` = the sum of related `jobs`.`jobvalue` records.”

--mySQL.com

Do calculated attributes (not fields!) violate relational theory and must be "normalized" out of them? Determining that requires foundation knowledge that is scarce in the industry, which has a poor and outdated understanding of the RDM.

Follow @DBDebunk Follow @ThePostWest

Note: To demonstrate the correctness and stability due to a sound theoretical foundation relative to the industry's fad-driven "cookbook" practices, I am re-publishing as "Oldies But Goodies" material from the old DBDebunk.com (2000-06), Judge for yourself how well my arguments hold up and whether the industry has progressed beyond the misconceptions those arguments were intended to dispel. I may revise, break into parts, and/or add comments and/or references. You can acquire foundation knowledge by checking out our POSTS, BOOKS, PAPERS, LINKS (or, even better, organize one of our on-site SEMINARS, which can be customized to specific needs).

ON DATA WAREHOUSES

(originally posted 11/10/2001)

“It is dawning on me that data warehouse techniques (as advocated by Ralph Kimball) are a response to perceived SQL DBMS performance weaknesses and nothing more. Dimensional modeling is a physical implementation design to deal with what may already be out of date performance tests to back up said design. I agree that where data warehouse logical design deviates from the relational data model there will be trouble down the road. I can attest to the high cost involved in maintaining a data warehouse. I am now questioning three purported benefits of using the current popular data warehouse design techniques.

Physical DBMS designs like the star schema produce faster more predictable query results than a normalized one (I realize normalization is a strictly logical concept but it does appear to have its direct physical mappings in current DBMS systems). Well, I am about to find out. We will be performing some benchmarks. What really are the query times performed on our OLTP system vs. a Star schema for a few relevant reports.

Data warehouses offload query processing from the OLTP system. This is true, but may not be necessary. One needs to thoroughly analyze the traffic on the OLTP system to see if offloading is necessary. We are looking into simply replicating the database (or part of it) for reporting purposes. Replication is far simpler to maintain than a data warehouse.

Data warehouse designs are simpler to understand than a relational one, thereby query construction is easier. I think this is more due to the fact that designers have had a bad habit of throwing up on a wall a full ER chart of their systems. Saying in effect Look how great I am, no one will ever understand this!  Creating ER diagrams of subsystems to describe important characteristics of a database can also be simple to understand.

Item #3 brings up a question. What do you think of OLAP tools such a Microsoft s Analysis Services? We have discovered you can use the tool without redesigning your database. We have seen some pretty fast query times if we take the option of allowing the tool to store data extracted from the production database into its own proprietary format. The opposition to its adoption here is the learning curve to master the MDX query language."

 

Fabian Pascal: The fact is that so-called "dimensional modeling" is logical, not physical modeling. Kimball does not present his "techniques" as just performance-maximization and, what is more, they won’t necessarily yield better performance. He also seems to believe, erroneously, that star-schemas are fully normalized designs. The real solution to performance problems is true RDBMSs with better implementations, not ad-hoc logical designs. Like so many, Kimball simply does not understand relational technology and confuses levels of representation.

I consider warehouses a regression to the good old days of application-biased files -- which we discarded for application-neutral and DBMSs, because they did not prove cost-effective. They are application-specific views of databases that are not derived via relational algebra from relational databases, but are rather non-relational SQL tables. The industry has a long and profitable history of recycling relabeled old failures, witness XML and graph throwbacks to hierarchic databases.

Aside from being necessary for soundness, fully normalized relational databases are the easiest to understand if (1) one thoroughly knows and understands the segment of reality to be represented in the database -- the business, that is -- and (2) data fundamentals. Arbitrary designs such as star-schemas are cope-outs, due to poor knowledge and understanding of the latter.

I am not familiar with the product, but I am generally wary of what vendors do.


Note on re-publication: See also Data Warehouses and the Logical-Physical Confusion.