How come rollbacks essential?
Could it be to avoid data (like data inside a SQL DB) from having an sporadic condition?
If that's the case, how comes the information "store" (the SQL DB or whatever) managed to get possible to begin with being inside a corrupt condition?
Exist data storage systems that do not need to have "rollbacks"?
Rollbacks are essential just in case associated with a type of errors showing up throughout database operational. They are able to really save your day just in case of database server crashes or perhaps a critical exception is tossed within an application that modifies items in DB. Whenever a significant DB operation is carried out (i.e. updates, card inserts, etc.) and the operation is damaged in the centre, it might be very difficult to trace which procedures were effective and use of DB after could be very complicated.
The "store" itself doesn't have a built-in mechanism for consistency control - this is why we use rollbacks and transactions. This is often regarded as a kind of 'live backup' mechanism.
You will find cases, when you really need place/update data in lots of related tables - should you did not have transactional logic, then any errors somewhere in middle of process might make data sporadic.
Simple example. Say you have to place both order header data into orders table and order lines into lines table. You place order header, read identity, start placing order lines - but this second place fails on unkown reasons. Only reliable method to get over this case would be to rollback first place - either clearly (whenever your link with db is alive) or unconditionally (when link is finished lower).