winscape 2346
 |
for a little topic drift ... there was an industry service that collected erep data (ras and error recording data) from large number of customer installations .... somewhat removed customer identifying information and published summaries on...
How versatile was this configuration base? Let me ask this in a different way. How did you evaluate a system that had non-IBM gear and-or connections to it? As sophisticated as that benchmark generator was, I don't see how it could predict the unknown based on unknowns. :-)
DEC never had that kind of internal equipment for any of our general hard-software releases. We were very lax at doing our regression tests (personnel problem that was allowed to continue for years). As a result, our stress and configuration testing (where the real code was exercised) was done by a careful selection of field test sites.
How good was it at predicting timesharing. On-line users as a population are the most fickle, unpredictiable and consistent uninteresting moment.
I don't know if anybody could manage to do a combination of yours and our approaches. It seems like this is another thing that evolves over time based on the choices made by customers and developers.
Now that I'm thinking about it, I bet the only data brought back inhouse from customer sites would have been a few facts gathered by Field Service. When people owned their gear, they also were leery about giving up their bits, too.
BAH