It's stated here:

http://www.ibm.com/developerworks/web/library/wa-dbdsgn2.html

Each table within the DB must have a brief history table, reflecting the whole good reputation for the main table. If records however table should be up-to-date, that old items in the record are first replicated towards the history table prior to the update is created. In exactly the same way, erased records within the primary table are replicated towards the history table prior to being erased in the primary one. A brief history tables also have the title from the corresponding primary one, however with _Hist appended.

In temporal db see here http://stackoverflow.com/questions/1514858/temporal-database-modeling-and-normalisation there is not another table so far as I realize.

Then when must i create another table or otherwise ?

History tables provide past (generally non-temporal) changes designed to the main database records by customers. This history is archival in character (i.e. utilized from time to time for historic reasons). The temporal information (once the change is made) is secondary in character.

A temporal database was created particularly to complete time queries against. The temporal details are primary in character, and stored online for fast retrieval. Another table isn't produced, unless of course archiving must also occur.

http://en.wikipedia.org/wiki/Temporal_database

What Robert stated theoretically - absolutely nothing to add.

Practically, temporal table versus. primary+hist table, has other impications.

For heavily maintained data (e.g. updates/removes greatly outnumber the card inserts), getting a historic (sometimes also known to as "audit" - because it is the primary mechanism to enforce audit trail of DB data) table enables keeping the primary table reasonably smaller in comparison to keeping the audit info within the primary table itself. This could have significant performance implications for chooses and card inserts around the primary table, especially considering index optimisation talked about below.

To top that off, the indices on hist/audit table don't need to be 100% just like primary table, meaning you are able to omit indices unnecessary for querying audit data from hist database (thus accelerating card inserts into audit table) and, the other way around, optimize what indices you will find to a particular audit queries you've (including ordering the table by timestamp via clustered index) without saddling the primary table with individuals indices which slow the information changes (and just in case of clustering promptly of update, clash with primary table's clustered index which means you usually can't get it clustered in temporal order).

A brief history table that's spoken of for the reason that developerworks article is really a table store the good reputation for the database (i.e. a brief history in our values about reality).

The type of history that you simply requested about for the reason that other thread holds our (current !) belief concerning the good reputation for reality.

Note the main difference. The 2 concur simply to the extent our past values about reality have indeed been correct. Which isn't necessarily 100%.

If you are using the previous as the latter, then you're in a sense presuming that that amount of concurrence is definitely 100%, i.e. that your past values about reality always by definition coincided with reality, i.e. you're presuming that it's impossible that you should have experienced any faulty belief about reality.

Tables that contain the good reputation for other tables can suit reasons of auditing. Tables that contain the good reputation for reality can suit the objective of any user that's interested for the reason that historic information.