Using an MD5 hash offers potential integration, performance, and security benefits.

When developing modern applications or enhancing legacy ones, an RPG or DB2 routine may need to apply an MD5 hash to application data. Fortunately, IBM i/OS offers a machine interface operation called CIPHER to make the task of implementing MD5 relatively simple.

What Is MD5, Anyway?

To answer this question, I'll just be lazy and quote from Wikipedia:

In cryptography, MD5 (Message-Digest algorithm 5) is a widely used cryptographic hash function with a 128-bit hash value. Specified in RFC 1321, MD5 has been employed in a wide variety of security applications, and is also commonly used to check the integrity of files.… An MD5 hash is typically expressed as a 32-digit hexadecimal number.

If you're unfamiliar with cryptography, your next question may be, "What is a cryptographic hash function?" Again Wikipedia gives a good intro:

A cryptographic hash function is a deterministic procedure that takes an arbitrary block of data and returns a fixed-size bit string, the (cryptographic) hash value, such that an accidental or intentional change to the data will change the hash value. The data to be encoded is often called the "message", and the hash value is sometimes called the message digest or simply digest.

The ideal cryptographic hash function has four main or significant properties:

• it is easy to compute the hash value for any given message,
• it is infeasible to find a message that has a given hash,
• it is infeasible to modify a message without hash being changed,
• it is infeasible to find two different messages with the same hash.

OK, so putting it all together, an MD5 hash is an algorithm that operates on a block of data and comes up with a unique 16-byte (128-bit) "signature" or "hash" that uniquely identifies the data. For the record, cryptographic algorithms are often referred to as "ciphers."

Hash vs. Encryption

It's important to note that a hash is different from encryption. A hash algorithm will simply come up with a unique identifier (i.e., hash) for the data; it does not alter the data itself. Further, the original data cannot be derived from its hash; hence, it is a "one-way" algorithm.

In contrast, encryption transforms data intended to be protected into an illegible format. Ideally, data that has been encrypted can't be decrypted into its original state without knowing the original algorithm and encryption key.

Generally, hash algorithms are used to protect or verify data integrity (such as verifying a zip file from a public FTP site hasn't been tampered with), whereas encryption is intended to scramble data for storage and later decryption and retrieval (such as a credit card or Social Security number).

Programming with MD5

Many languages and database systems have the ability to calculate an MD5 hash function (usually via an API), including the .NET family of languages (C#,VB, etc.), Java, PHP, SQL Server, Oracle, and even the IBM i languages (RPG, COBOL, and C.)

What Can I Do with an MD5 Hash on the IBM i?

Having an MD5 hash available has the following potential applications:

Create Hashes to Uniquely Identify Data

If you create a file on the IFS and need to verify that its contents have not been tampered with, you can calculate and store a unique hash for the file. The same application applies to database data.

MD5 hashes can be used to compare query result sets. For example, if I rewrite a query for performance reasons, I often want to make sure that the new result set matches the original result set. If I have a large result set created by Query A involving character data expressions, and I want to compare the results (in total) to Query B, I can let a hash algorithm verify that the data is the same between the two result sets both in data values and the order of the rows. If anything is different between the two sets, the hashes will be different.

Store Hashes Instead of Private Data

If you have an application that requires a custom-written user security system (instead of using i/OS user profiles) such as a Web application with a large number of users, when a user first defines his or her password, you can simply compute and store the password hash instead of storing the password itself. Every time a user attempts to log in, the login process only needs to re-compute the hash and check to see if the hash of the login password matches the original value stored in the database. Not storing the password at all prevents the potential security exposure of stolen passwords. Remember, the original password cannot be reconstructed from its hash.

Hashes Can Reduce Storage and Improve Performance

If you have an application that has to do a unique value lookup on a large character column, then an MD5 hash will allow you to save index storage and potentially increase performance. For example, if you have a CHAR(100) indexed column (or even a group of columns), you can create a unique hash and store the value in a BINARY(16) column. The lookup can be done against the hash value instead of the data columns. Building an index on a 16-byte column is preferable to a 100-byte column!

Currently on DB2 for i, you can't use a user-defined function in an index expression. Therefore, you need to store the hash value in a separate column so that the hash value can be indexed. Keep in mind, this index built on a hash column will be useful only for "equal value" lookups; SQL LIKE or RPG SETLL operations will not work. Further, if the index has to be updated often, it could negatively affect performance because the binary hash values will basically be random and can cause additional index maintenance overhead.

Provide Compatibility with Other Languages and Environments

Here are a couple of examples:

• An RPG program needs to verify an MD5 password hash that's stored in Oracle.
• You're converting a SQL Server stored procedure to DB2 for i and need to replicate the functionality of the HashBytes function. You can now write your own limited HashBytes equivalent using a DB2 user-defined function.

An RPG Service Program Example

As a sample, RPG service program MD5R contains subprocedures MD5Hash_BINARY and MD5Hash_HEX, which will create an MD5 hash on a string value and return the hash value as either a 128-bit binary value or a 32-character hex value.

Both subprocedures are very small. The CIPHER machine interface (MI) operation does all the hard work of calculating the hash. All the RPG program needs to do is correctly define the "controls" (aka input or parameters) and be able to retrieve and interpret the output of CIPHER.

These subprocedures can be called from another ILE program or be invoked by DB2 as a user-defined function (UDF). Sample DB2 for i CREATE FUNCTION statements are included in the source header documentation to demonstrate how to make these subprocedures available to DB2. For example, the following statement shows how UDF md5hash_hex can be invoked in a query:

SELECT md5hash_hex('mypassword') FROM qsqptabl

The result is a 32-character hex string: 34819D7BEEABB9260A5C854BC85B3E44.

Please note that the sample DB2 user-defined functions have their input parameters defined with CCSID 1208 (UTF-8). This means that the input data will be translated by DB2 to 8-bit Unicode (similar to ASCII in many respects) before passing the data to the RPG program so that the function will perform similarly to hash functions found on non-EBCDIC systems. If you want the hash applied against the EBCDIC data, then the parameter CCSID should be changed to the appropriate value (CCSID 37 is US EBCDIC). 

Likewise, if you want an ILE program to operate in a mode compatible with other systems, such as a Java program or SQL Server, then the EBCDIC data will need to be converted to an ASCII code page or Unicode first before calling the hash subprocedure. The iconv API offers one possible mechanism to accomplish this conversion within an RPG program.

One special "control" or "parameter" of the CIPHER operation to note is "sequence." Although not demonstrated in the sample program, specifying the sequence correctly can allow CIPHER to build a hash on multiple non-contiguous values that are passed across multiple calls to the MI. This means you can create a single hash from three independent pieces of data, such as name, address, and phone. Implementing this feature in a DB2 function (using the SCRATCHPAD feature of external user-defined functions) allows you to create a "running" hash on all the values in a column.

A New Tool for the Box

Using an MD5 hash opens up all kinds of integration, cross-database compatibility, performance, and security options for your applications. Take advantage of the i/OS built-in APIs to give your applications these advanced capabilities.

as/400, os/400, iseries, system i, i5/os, ibm i, power systems, 6.1, 7.1, V7,

Also Read

Programming in ILE RPG - Defining Data Structures, Part I

H and F Position-less Specs

Practical RPG: AI and IBM i