As the name implies, SQL injection vulnerabilities allow an attacker to inject (or execute) SQL commands within an application. It is one of the most wide spread and dangerous application vulnerability. The CLASP project provides a good overview of SQL injection.
Append the output of one or more other statements
Blind SQL injection
Insert, update and delete
Alter .. in case of a dynamic query.
Query other databases with @database link
The reason that SQL injection is possible, is that the SQL statement is being generated from a language or enviroment that isn't working with SQL natively. So it does not know the consequences. The same goes for the database itself. As long as the SQL statement is syntactally correct and user that performs it has the rights, the database will execute it. There is no way for both side to determine what the impact would be. When trying to figure that out your essentially blacklisting and any new variant will be bypass your predetermined.
Another solution often found is to obfuscate the table and columns. Now this is not going to work as one of the 13 rules that describe an Relational Database is that the metadata or catalog describing the database is stored in the database. For instance a query on ALL_TAB_COLUMNS would return me all the availeble tables, it columns and the types. So we need to make the understand eachother better. We can do this by seperating the data from the logic.
As with all languages and databases, the seperation of the SQL logic and the SQL data is the only way to prevent SQL injection. These are placeholders we can use to tell the database these parts of the query are dynamic and the rest is hardcoded. The communication is split up into multiple packages
But this is how a database wants you to communicate with it. Whenever a query passed to an Oracle database for execution, the first thing the RDMS will do is make a hashvalue of the query. Next it will to find the best way to collect the data that is needed and stores that in it's memory. After he is done with that it will selects the data and pass it on to the requester. A general rule is that if a query is responds within several seconds it was probaly spending more time in optimisation than the actual fetching.
When you are hardparsing each query, which is essentially the same but using diffrent parameters or values to match, this will result in a diffrent hash value for each statement. Thus the DBMS has to do it's work again. When you are softparsing (ie. using bind variables) the first time you execute the query it will take a fraction longer than the hardparsed version as it needs to send more packages to the database. But each time thereafter the database can skip that bit and just fetch what is needed. And if it is still in the buffercache even give you the results immediatly.
This is probably the only attackvector that when it is correctly fixed, it will also give you an performance improvement :)