When loading info into postgres, we have performance issues as the numeric and text loading docks get "big" (one is just under a billion rows and showing stress).
We should be able to take a very comprehensive approach to this. First, we will partition the data by month, however as each check is monotonically increasing in time, we can rotate out checks into the partition tables as the 5 minute segment is closed out. This means we can keep an "live" loading dock table that basically only had about 5 minutes worth of data for each check in the system.
