Product usage computation, interfaces, and behaviors

This kind of usage data is a great fit for a time series database. I built Hyprcubd to handle things like this and we actually "dog-food" our own product as well. As queries come in, we use an internal Hyprcubd table to track queries and usage. Check us out here https://www.hyprcubd.com/

• We have a dedicated Usage area that shows users how much they have consumed for the month. Metrics in your case would be counts for emails and users and yes we run simple count queries for those. For GB stored it is a little more tricky. You have to calculate the average over a specified time period such as a day, hour, etc then sum it across all your items. For us, we calculate usage per table broken down by the minute.

• We run count queries because Hyprcubd is designed to handle this. Depending on your scale, you might have to do rollups in another database rather than operating on raw data.

• We haven't set limits yet, however I would prefer to notify the user if they reach their limit or even go over rather than blocking them. I would rather not want to upset a paying customer that forgot to pay than to block them and interrupt their service. At that point, communicate with the customer.