Re: CODASYL-like databases

From: Ken North <>
Date: Wed, 02 Apr 2008 08:14:34 GMT
Message-ID: <KfHIj.31076$>

The CODASYL standard was a network model database. Conceptually it was similar to an on-disk representation of a doubly-linked list, with an additional pointer from a set member to its owner. So you could navigate through sets by following an next, previous or owner pointer.

CODASYL databases put a premium on database design because there was nothing comparable to ALTER TABLE. Best practice called for diligence in determining what the contents of sets were and the data (type, precision). The reason was that making changes was arduous.

For example, if you had a database with name address information, such as subscribers and customers, and you decided to add a second telephone number:

  1. You recompiled your schema and sub-schemas (embedded in application source code)
  2. Then you had to reload the entire database (often not the best way to spend a weekend).

CODASYL-type database were fast but inflexible. And broken pointers could cause anomalies on updates and insertions.

The group that defined the CODASYL standard did a survey of existing database systems in the early phase of its work. Here's what they reported in that 1968 survey:

  • Ken North ===========

"Troels Arvin" <> wrote in message news:fsslgl$320$
> Hello,
> RDBMSes are sometimes described as a reaction against network-/
> hierarchical databases. I believe that I've read that CODASYL-like
> DBMSes
> resulted in databases which were very hard to maintain.
> Does someone know of more concrete descriptions of what problems the
> old
> CODASYL-like databases resulted in?
> --
> /home/troels/.signature
Received on Wed Apr 02 2008 - 10:14:34 CEST

Original text of this message