File Transfer Protocol (FTP): Control vs Data Connections and How FTP Works

File Transfer Protocol (FTP ) works by using two separate TCP connections between a client and server: a control connection for commands and replies, and a data connection for the actual file contents. This separation allows the client to exchange control information (e.g., login, directory listing commands) while the transfer itself happens over a distinct channel.

Below is a neat high-level diagram showing the typical behavior for a single file transfer where the client sends FTP commands on the control channel, and the server opens/receives a new connection for the data channel.

Key ideas

• Control connection carries commands like USER, PASS, RETR, and reply codes like 150 and 226.
• Data connection is created separately for each transfer (e.g., for RETR or STOR).
• FTP uses a command-reply model on the control channel; the data channel exists only when a transfer is taking place.

Control Connection in FTP

The control connection is the “signaling channel” for FTP. On this connection, FTP uses an ordered exchange of commands and numeric replies, allowing the server to report the state of the session and of transfers (e.g., permission checks, errors, and readiness).

Common properties:

• It is established once per FTP session (typically client connects and stays connected during the session).
• It carries:
  • FTP commands such as USER, PASS, RETR, STOR.
  • FTP reply codes that indicate progress and results.
• It coordinates creation of the data connection, but does not carry file bytes.

Control connection lifecycle (conceptual)

Data Connection in FTP

The data connection is used only when transferring data such as file contents or directory listings. Unlike the control channel, the data connection is typically created per transfer (or at least per transaction), which enables the protocol to keep control signaling responsive while the bulk transfer happens.

Conceptually:

• The control channel tells both sides a transfer is about to happen.
• The data connection provides the bulk pipe for bytes.

Why separate connections?

Separating signaling from payload helps FTP remain structured:

• Control messages (commands/replies) remain available during transfer coordination.
• The bulk transfer uses a connection suited for streaming file bytes.

Active vs Passive Modes (How the data connection is created)

FTP must create a new TCP connection for the data channel. The method used depends on whether the client uses active or passive mode.

Active mode (high level)

• Client tells the server where to connect for the data channel (e.g., via PORT command).
• Server then connects back to the client’s indicated address/port to start the data transfer.

Passive mode (high level)

• Client sends PASV and the server responds with an address/port where it is listening for data.
• Client then connects to the server’s provided address/port for the data transfer.

Visual comparison