Breaking a large table into small ones?

Hello DB experts,

I would like to know how many rows before people break a table containing large rows down to many horizontal tables you recommend? For examples, a phone company that covers entire U.S., do they use one table for the entire nation or one table for each state? Or The U.S. Social Security Department, do they use one table for entire population or one table for each state? Or do they break it down further like one table for each city?

Thanks,
John

They would have to break it down, say by state, I dont know of a database that has a field type, with UNLIMITED characters, or rows, databases like mysql max out at 64345 rows or something like that,

correct me if im wrong tho

Really, so low? I would say that you don't determine how many tables to use by the number of rows, but you would base the amount of normalization on how many rows and what ways the data might be looked up. For example, looking up people by state is faster if you are searching a foreign key corresponding to a state in a state table, rather than a text search for the state name or abbr.

Well, I asked same question at other 2 forums, they say it depends on how size of each row, the larger the row, the less rows you have and vice versa. Size of a table max out at 2G or 4G depend on OS.
http://www.mysql.com/doc/en/Table_size.html

Correct me if I'm wrong but in my understanding there is sometimes a fixed set amount of rows that are returned when querying from php or asp.. At least with some type of db's

For example, Ms Access has a limit of 10.000 rows standard to be returned at all times. (No top10000 in the sql statement) You can change this setting higher but still strange if you don't know and you're stuck with 10.000 rows while there are more.

Apart from the database size in the rows versus depth discussion... the width of the line will speed up database performance but naturally the depth of the table will be lower..

Thus in response to johnn and the location on the website pointed out. (thx for that one... handy table) I would say that using a lot of relational tables or databases all linked will not only increase speed of your database but allow you to have greater depth.

By the way, does MySql supports OLAP designs?