On the morning meeting. Steve, Wayne, Dustin, Eric, Alan, and I were in today. Wayne and Eric were talking query caching techniques and switching between aggregate sums and totals and non aggregated (simple transactional data). The conversation kinda went along the levels of watchers, feeders, hooks, triggers, etc. Eric was saying are we going to be building new aggregates based on the old design or are we going forward with a new design and then build aggregates off of the new design. Pretty deep data model changes. What are the impacts to those changes?

One of the questions came up about testing impacts and how does that work. Are we fixing legacy and/or being pro active and moving forward. One of the questions was the po/invoice line items table. It currently has both PO lines (inbound inventory) and invoice lines (outbound inventory) in the same table. The reason goes clear back to location specific levels of inventory tracking. If no locations were ever needed, we would hold thing (line items) in the different tables. However, if you add an unknown number of locations, that complicates the data model and it get hard to be able to roll the inventory levels both backwards (going back in history) and/or seeing current levels.

Going on to a different topic, Wayne and Eric were talking about a testing server and even stress testing certain queries and/or database tables. They were talking about caching queries and how often do we reboot these servers. Because we use Adobe ColdFusion, we've had to implement a nightly reboot to help flush memory. As Wayne gets into the servers and monitoring things, we are seeing that the servers are working pretty hard by the end of the day and then fully refreshing things based off of reboots. That is both good and bad. We are on purpose flushing the cache and the memory. However, sometimes a cache and/or an item held in memory saves tons of time. We still need to compare the info and see what the differences and advantages are. Like Alan said - we break stuff and then we fix it.