A SQL injection attack is a code injection technique where the user input is interrupted as SQL commands rather than standard data while processing the SQL query.
This vulnerability is exploited by an attacker to access, modify and delete data from the database.
By leveraging SQL Injection, an attacker will be able to bypass authentication and access unauthenticated data from databases. SQL injection vulnerability will potentially allow the attacker to escalate to more damaging attacks inside the network where the application is hosted.
Based on different strategies attackers use to exploit SQL injection vulnerabilities, they can be categorised as:
For in-band SQL injection, the attacker uses the same communication channel to execute attacks and retrieve information from the application.
In-band SQL injection is easy to exploit and a commonly used injection technique among the attacker community.
The In-band SQL injection vulnerability can be further classified into four:
Blind SQL injection is an injection attack where no actual data is transferred between the attacker and the application.
It is also known as Inferential SQL injection. The attacker cannot see if the inserted commands are executed or not. This is why this attack is called a blind attack.
The attacker in this attack tries to construct a separate database inside the application. The attacker determines his success by reading the response from the application.
The following types of injection attacks come under blind SQL injection:
Out-of-band SQL injection is an injection attack that is rarely used by an attacker. The possibility of this attack depends on the features enabled in the database of the server.
This attack is used when executing commands and getting information can be done through a single communication channel.
Out-of-band SQL injection is used as an alternative for time-based blind SQL injection. This attack is also used if the server is not giving stable responses. If a server cannot provide a proper response, it can be easily stalled by sending requests from many different systems.
Depending on the backend database configuration, access privileges and the operating system, an attacker can mount one or more of the following type of attacks:
Beagle Security recommends implementing the following defence measures:
Stored procedures are not always trustworthy in all types of SQL injection.
When appropriately implemented a stored procedure has the same effect of a parametrised query. A stored procedure language requires the developer to build SQL statements with parameters.
A SQL query for a stored procedure is defined and stored in the database itself, which is then invoked from the application.
In a prepared statement with a parameterised query, the developer has to define all the SQL query and pass in each parameter to the query later.
A prepared statement helps the database to differentiate between code and data, regardless of the input. This statement prevents an attacker from changing the intent of a query even if SQL commands are inserted within the parameters.
There are many illegal locations for the use of bind variables such as the names of tables or columns and sort order indicator. Whitelisting input validation is the most appropriate defence to tackle this situation.
The name of tables or columns is usually extracted from code and not from user parameters. But, bind variables use parameter values. The parameter values are to be mapped to the legal expected table or column names to make sure invalid user input doesn’t get executed in the query.
For handling sort order, the best method is to convert the user-supplied input into boolean, and then use that boolean to select the safe value to append to the query. Converting user-supplied input into boolean is an essential step in creating dynamic queries.
Escaping all user-supplied input technique is used when all other injection prevention techniques are unusable.
Implementing this mitigation is weak compared to the other defences. This fix will not guarantee prevention of a SQL injection in all situations.
Beagle Security recommends this technique to retrofit legacy code when there is ineffective input validation. This technique supports one or more character escaping schemes in each DBMS that is specific to certain kinds of queries.