This article is dedicated to the task of securing MySQL client-server connection using functionality provided by the Secure Shell (SSH) protocol. To be exact, the SSH tunneling concept is utilized. We will review the steps needed to build secure MySQL client applications and implement a sample one ourselves.

MySQL traffic is not the only kind of data that can be tunneled by the Secure Shell. SSH can be used to secure any application-layer TCP-based protocol, such as HTTP, SMTP and POP3. If your application needs to secure such a protocol by tunneling it through a protected SSH connection, this article will be useful to you.

Background

Let’s imagine that we are developing an enterprise application that needs to send requests to a number of SQL servers all over the world and get responses from them (let’s imagine that it’s a super-powerful bank system that stores information about millions of accounts).

All the data between the application and SQL servers are transferred via the Internet “as is”. As most protocols used by SQL servers do not provide data integrity and confidentiality (and those that do, do it in a quite nontransparent way), all the transferred requests and responses may (and be sure, they will!) become visible to a passive adversary. An active adversary can cause much more serious problems – he can alter the data and no one will detect it!

SSH (Secure Shell) is a protocol that may help in solving this problem. One of its outstanding features is its ability to tunnel different types of connections through a single, confident and integrity-protected connection.

Now you do not have to worry about securing the data transferred over the Internet – SSH will handle this for you. In particular, SSH will take care of the following security aspects:

Strong data encryption according to the latest industry-standard algorithms (AES, Twofish)
Authentication of both client and server computers
Data integrity protection
Stability with regard to different kinds of network attacks
Compression of the data being tunneled
Complete independence of the operating system and network specifics
Tunneling (or forwarding) works in the following way:

SSH client opens a listening port on some local network interface and tells the SSH server that he wishes to forward all connections accepted on this port to some remote host.
When another connection is accepted on the listening port, the SSH client informs the SSH server about this fact and they together establish a logical tunnel for it. At the same time, the SSH server establishes a new TCP connection to the remote host agreed upon in step 1.
The SSH client encrypts all the data it receives from the accepted connection and sends it to the SSH server. The SSH server decrypts the data received from the SSH client and sends it to the remote host.

Please note, that the SSH client acts as a TCP server for the connections it accepts, and the SSH server acts as a TCP client for the connections it establishes to the remote host.

A single SSH connection can tunnel as many application layer connections as needed. This means that you can defend your server by moving all the listening ports (e.g., database and application server ports) to a local network, leaving only the SSH port open. It is much easier to take care of a single port, rather than a dozen different listening ports.

Into the Fire!

Let’s develop a small application that illustrates the use of SSH forwarding capabilities. We will consider an important task of securing a connection between a MySQL client application and a MySQL server. Imagine that we need to get information from the database server, which is located a thousand miles away from us, in a secure way.

SecureMySQLClient is the application we are planning to implement. It includes the following modules:
SSH client-side module with forwarding capabilities
MySQL client-side module
User interface for configuring application settings and displaying query results.

The SSH server runs in a remote network and is visible from the Internet. The database (MySQL) server runs in the same network as the SSH server and may not be visible from the Internet.

The process of performing secure data exchange between SecureMySQLClient and the Database server goes as follows:

The SSH client module negotiates a secure connection to the SSH server and establishes forwarding from some local port to the remote MySQL server.
The MySQL client module connects to the listening port opened by the SSH client module.
The SSH client and server set up a logical tunnel for the accepted connection.
The MySQL client sends SELECT to the port opened by the SSH client module, which encrypts it and sends it to the SSH server. The SSH server decrypts the request and sends it to the MySQL server.
The SSH server receives a response from the MySQL server, encrypts it and sends it back to the SSH client, which decrypts it and passes it to the MySQL client module.

Looks too complex? Implementing this is easier than you think.So, let’s go and do it.

We will need the following products installed on the computer before creating the application:

Microsoft Visual Studio .NET 2003, 2005 or 2008.
EldoS SecureBlackbox (.NET edition). Can be downloaded from
http://www.eldos.com/sbbdev/download.php.
MySQL .NET Connector. Can be downloaded from
http://www.mysql.com/products/connector/net/.

Let’s now open Microsoft Visual Studio .NET (we will use the 2005 version) and try to build such an application from scratch.

After the GUI design has been finished, we can go on with the business logic code itself. First, adding references to the following assemblies to our project:

SecureBlackbox
SecureBlackbox.PKI (only in SecureBlackbox 5. SecureBlackbox 6 doesn’t have this assembly)
SecureBlackbox.SSHClient
SecureBlackbox.SSHCommon
MySql.Data

SSHForwarding notifies us about certain situations via its events, so we need to create handlers for some of them:

OnAuthenticationSuccess – Is fired when the client authentication process has been completed.
OnAuthenticationFailed – Is fired if the client was unable to authenticate using particular authentication method. In general, this does not mean that the authentication process completely failed – the client may try several authentication methods consequently and one of them may succeed.
OnError – Is fired if some protocol error occurs during the session. Usually this leads to a connection closure. The exact error can be detected via the error code passed to it.
OnKeyValidate – Is used to pass the received server key to the application. Please note that incorrect handling of this event may result in a serious security breach. The handler of this event should verify that the passed key corresponds to the remote server (and warn the user if it does not). If the key is valid, the handler should set the Validate parameter to true. The sample does not perform key checkup for the sake of simplicity.
OnOpen – Is fired when the SSH connection is established and the component is ready to tunnel data. We will use the handler of this event to kick the MySQL client component.
OnClose – Is fired when the SSH connection is closed.
OnConnectionOpen – Is fired when a new tunnel is created. The corresponding tunneled connection object is passed as parameter.
OnConnectionClose – Is fired when an existing tunnel is closed.

Implementing two core methods, SetupSSHConnection() and RunQuery(). The first one initializes the SSHForwarding object and establishes an SSH session to the remote server by calling its Open() method, and the second one sends the query to the MySQL server.
The code of the SetupSSHConnection() method is pretty simple:
private void SetupSSHConnection()
{

Pin It on Pinterest