>> You said that using an EntityID doesn't scale well.
>> I'm just wondering how the big boys do it?
They spend a lot of money.
But I imagine they also segment their databases where possible. Single databases that span multiple servers are complicated to set up, and expensive (from a human point of view) to keep going.
On the other hand if you are creating distinct data sets where the data is naturally segmented than it becomes very easy to add more servers, put different companies on different servers, make backups, and so on. Unless you absolutely need all the data squished together in the same DB, I wouldn't go that route myself. Splitting data across databases seems to make a lot more sense.
>> folders/paths for TPS files or databases in SQL
>> which would have a faster response to the user?
response time for the user is similar in both cases - with the usual endless caveats on size, reliability and so on. Both are fast enough, so I think speed should not be the determining factor here in the short term. It's also fairly straight-forward to change from one to the other, so you're not locking yourself in whichever way you go. (TPS can be helpful when you're learning, if only because it's a lot easier to send me examples.)
Cheers
Bruce