Re: InfoWorld Test Plan

In article <D806y6.KA6_at_postoffice.ptd.net> edowgial_at_postoffice.ptd.net (Edward Anthony Dowgiallo) writes:
>From: edowgial_at_postoffice.ptd.net (Edward Anthony Dowgiallo)
>Subject: InfoWorld Test Plan
>Date: Wed, 3 May 1995 13:03:42 GMT
 

>InfoWorld is developing a new test plan for evaluating departmental SQL
>servers. This would cover systems like Gupta SQLBase, Ocelot, Oracle WG,
>Watcom, and XDB.
 

>I'd like to hear from you what factors you consider most important in a
>departmental server. Is connectivity to larger engines critical or is
>alright for these guys to be strcitly standalone? Do you want to see the
>same high end development features like declarative referential
>integrity, stored procedures, etc? Is roll back/roll forward recovery
>required? Do you want to see native connectivity from 3rd party
>development tools or is ODBC OK?
 

>Please give us some input on what you'd like to see tested.
 

>Ed Dowgiallo
>InfoWorld Review Board

Ed -

I think that referential integrity is pretty basic at this point, so it should be included. Same with recovery, although there might be some interesting differences in how easy recovery is, or if you can recover from a damaged database.

One thing you have not mentioned, and is generally skipped, is how multi-user issues are handled. I think that by the classification of 'departmental' and the feature sets you are considering that the servers will be classic client/server backends. By this I mean heavy read access and long running transactions.

Many people do not understand how different locking strategies and isolation levels can dramatically impact the success or failure of an database application. It would be very useful to explore the differences between these products in this area.

• Rick Greenwald Performance Computing

Rick Greenwald
Performance Computing
greenie_at_interaccess.com Received on Wed May 03 1995 - 00:00:00 CEST