The project I've been working on since May has involved replacing one database load system with another. The old way involves programs going thru a translation layer to produce flat files that are copied to the remote host and loaded to the target. The new way involves a "pull" directly from the remote host. The designer of the new system had estimated that it should be faster, besides simplifying the implementation.
Yesterday we had our first chance to compare "old way" and "new way" in a meaningful manner. It's taken so long because the system was so complicated and we had so few resources.
The new way takes the same time to run as the old way. It was a huge letdown.
When you think about it, we're taking the same megabytes of data, moving them across the same wire, and dropping them into the same database. "Captain, I canna change the laws of physics." Just shows the old system wasn't that inefficient after all. Which we may have known if we'd spent more time up front, but that was another resource we didn't have.
I've got another 2 weeks to squeeze more juice out of this. Wish me luck.