Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!dptg!ulysses!andante!alice!debra
From: debra@alice.UUCP (Paul De Bra)
Newsgroups: comp.databases
Subject: Re: Ingres and referential integrities
Keywords: Ingres referential integrities
Message-ID: <9653@alice.UUCP>
Date: 21 Jul 89 13:12:09 GMT
References: <3081@rti.UUCP>
Reply-To: debra@alice.UUCP ()
Distribution: comp
Organization: AT&T, Bell Labs
Lines: 25

In article <3081@rti.UUCP> slc@rti.UUCP (Lynn Cohen) writes:
>I'm curious as to why Ingres doesn't provide referential integrities.
>I'm aware that you can build in referential constraints through QBF
>and VIFRED but not as part of the data definition. Does anyone know
>why Ingres chose not to provide this part of the relational model?
>

My guess would be performance.

I once attended a talk by an IBM person (a techie, not a salesman)
who explained that when moving from their hierarchical database system
to the relational one there still was an (average) performance
loss of about a factor 3. This was even without referential integrity.

The reason he gave for having difficulties with referential integrity
was that in the hierarchical databases referential integrity comes
natural, it is almost built into the model. With a relational database
checking for referential integrity involves real (computational) work.

Paul.

-- 
------------------------------------------------------
|debra@research.att.com   | uunet!research!debra     |
------------------------------------------------------