Archives of the TeradataForum

Message Posted: Tue, 20 Jun 2000 @ 15:55:15 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

The Teradata support staff here at Boeing came out of scientific parallel computing (Cray, CDC) where all problems reported to the vendor (except for security issues) were stored in a common database that was available to all. We were frankly horrified at the secretive nature of the NCR problem reporting process. NCR claimed that their customer base would not tolerate problem sharing, i.e. Walmart wouldn't want Kmart to know 1) that they were having problems, and 2) what kind of SQL was causing the problem.

We demonstrated the support systems from other vendors to NCR during the Support Link development process, and we like to think that some good came of it. Fixing the problems would mean completely overhauling the incident reporting systems. For each incident you need:

* formal public problem reporting area,

* formal vendor response area,

* formal rebuttal areas,

* ability to log an interest in a problem reported by someone else,

* status and tracking information,

* automatic notification of changes in public fields, and

* hidden problem discussion area for the vendor.

People catch on pretty quickly that anything they put in the public area becomes a matter of public record, and they usually manage to remain cordial throughout. The last section is important for the developers so that they can freely debate the problem without worrying if they're wrong.

When I started working on CDC systems, they were mailing microfiche to the sites on a monthly or quarterly basis. Later they converted to an email-based list server, but today you'd do this on the web. Of course, NCR would need to find a good database engine somewhere... :)

	<Prev	Next>		<<First	<Prev	Next>	Last>>