Since mechanisms likeTelnet andRemote Shell are designed to access and operate remote computers, sending theauthentication tokens (e.g.username andpassword) for this access to these computers across apublic network in an unsecured way poses a great risk of third parties obtaining the password and achieving the same level of access to the remote system as the telnet user. Secure Shell mitigates this risk through the use ofencryption mechanisms that are intended to hide the contents of the transmission from an observer, even if the observer has access to the entiredata stream.[2]
Finnish computer scientist Tatu Ylönen designed SSH in 1995 and provided an implementation in the form of two commands,ssh andslogin, as secure replacements forrsh andrlogin, respectively. Subsequent development of the protocol suite proceeded in several developer groups, producing several variants of implementation. The protocol specification distinguishes two major versions, referred to as SSH-1 and SSH-2. The most commonly implemented software stack isOpenSSH, released in 1999 as open-source software by theOpenBSD developers. Implementations are distributed for all types of operating systems in common use, includingembedded systems.
SSH applications are based on aclient–server architecture, connecting anSSH client instance with anSSH server.[3] SSH operates as a layered protocol suite comprising three principal hierarchical components: thetransport layer provides server authentication, confidentiality, and integrity; theuser authentication protocol validates the user to the server; and theconnection protocol multiplexes the encrypted tunnel into multiple logical communication channels.[1]
SSH may be used in several methodologies. In the simplest manner, both ends of a communication channel use automatically generated public-private key pairs to encrypt a network connection, and then use apassword to authenticate the user.
When the public-private key pair is generated by the user manually, the authentication is essentially performed when the key pair is created, and a session may then be opened automatically without a password prompt. In this scenario, the public key is placed on all computers that must allow access to the owner of the matching private key, which the owner keeps private. While authentication is based on the private key, the key is never transferred through the network during authentication. SSH only verifies that the same person offering the public key also owns the matching private key.
In all versions of SSH, it is important to verify unknownpublic keys, i.e.,associate the public keys with identities, before accepting them as valid. Accepting an attacker's public key without validation will authorize an unauthorized attacker as a valid user.
OnUnix-like systems, the list of authorized public keys is typically stored in the home directory of the user that is allowed to log in remotely, in the file~/.ssh/authorized_keys.[4] This file is respected by SSH only if it is not writable by anything apart from the owner and root. When the public key is present on the remote end, and the matching private key is present on the local end, typing in the password is no longer required. However, for additional security, the private key itself can be locked with a passphrase.
The private key can also be looked for in standard places, and its full path can be specified as a command-line setting (the option-i for ssh). Thessh-keygen utility produces the public and private keys, always in pairs.
File managers for UNIX-like systems (e.g.,Konqueror) can use theFISH protocol to provide a split-pane GUI with drag-and-drop. The open source Windows programWinSCP[6] provides similar file management (synchronization, copy, remote delete) capability usingPuTTY as a back-end. Both WinSCP[7] and PuTTY[8] are available packaged to run directly off a USB drive, without requiring installation on the client machine. Crostini onChromeOS comes with OpenSSH by default. Setting up an SSH server in Windows typically involves enabling a feature in the Settings app.
SSH is important incloud computing to solve connectivity problems, avoiding the security issues of exposing a cloud-based virtual machine directly on the Internet. An SSH tunnel can provide a secure path over the Internet, through a firewall to a virtual machine.[9]
TheIANA has assignedTCPport 22,UDP port 22 andSCTP port 22 for this protocol.[10] IANA had listed the standard TCP port 22 for SSH servers as one of thewell-known ports as early as 2001.[11] SSH can also be run usingSCTP rather than TCP as the connection-oriented transport layer protocol.[12]
In 1995,Tatu Ylönen, a researcher atHelsinki University of Technology in Finland designed the first version of the protocol (now calledSSH-1) prompted by a password-sniffing attack at hisuniversity network.[13] The goal of SSH was to replace the earlierrlogin,TELNET,FTP[14] andrsh protocols, which did not provide strong authentication nor guarantee confidentiality. He chose the port number 22 because it is betweentelnet (port 23) andftp (port 21).[15]
Ylönen released his implementation asfreeware in July 1995, and the tool quickly gained in popularity. Towards the end of 1995, the SSH user base had grown to 20,000 users in fifty countries.[16]
In December 1995, Ylönen founded SSH Communications Security to market and develop SSH. The original version of the SSH software used various pieces offree software, such asGNU libgmp, but later versions released by SSH Communications Security evolved into increasinglyproprietary software.
It was estimated that by 2000, the number of users had grown to 2 million.[17]
In 2006, after being discussed in a working group named "secsh",[18] a revised version of the SSH protocol,SSH-2 was adopted as a standard.[19] This version offers improved security and new features, but is not compatible with SSH-1. For example, it introduces new key-exchange mechanisms likeDiffie–Hellman key exchange, improveddata integrity checking viamessage authentication codes likeMD5 orSHA-1, which can be negotiated between client and server. SSH-2 also adds stronger encryption methods likeAES which eventually replaced weaker and compromised ciphers from the previous standard like3DES.[20][21][19] New features of SSH-2 include the ability to run any number ofshell sessions over a single SSH connection.[22] Due to SSH-2's superiority and popularity over SSH-1, some implementations such as libssh (v0.8.0+),[23]Lsh[24] andDropbear[25] eventually supported only the SSH-2 protocol.
In January 2006, well after version 2.1 was established,RFC4253 specified that an SSH server supporting 2.0 as well as prior versions should identify its protocol version as 1.99.[26] This version number does not reflect a historical software revision, but a method to identifybackward compatibility.
In 1999, developers, desiring availability of a free software version, restarted software development from the 1.2.12 release of the original SSH program, which was the last released under anopen source license.[27] This served as a code base for Björn Grönvall's OSSH software.[28] Shortly thereafter,OpenBSD developersforked Grönvall's code and createdOpenSSH, which shipped with Release 2.6 of OpenBSD. From this version, a "portability" branch was formed to port OpenSSH to other operating systems.[29]
As of 2005[update],OpenSSH was the single most popular SSH implementation, being the default version in a large number of operating system distributions. OSSH, meanwhile, has become obsolete.[30] OpenSSH continues to be maintained and supports the SSH-2 protocol, having expunged SSH-1 support from the codebase in the OpenSSH 7.6 release.
In 2023, an alternative to traditional SSH was proposed under the nameSSH3[31][32][33] by PhD student François Michel and Professor Olivier Bonaventure and its code has been made open source.[34] This new version implements the original SSH Connection Protocol but operates on top ofHTTP/3, which runs onQUIC. It offers multiple features such as:
Faster session establishment, reducing the number ofRound-trip delays from 5-7 to 3.
High security: while SSHv2 relies on its own protocols, SSH3 leveragesTLS 1.3,QUIC, andHTTP.
UDP port forwarding
X.509 certificates
OpenID Connect
However, the name SSH3 is under discussion, and the project aims to rename itself to a more suitable name.[35] The discussion stems from the fact that this new implementation significantly revises the SSH protocol, suggesting it should not be called SSH3.
Example of tunneling anX11 application over SSH: the user 'josh' has "SSHed" from the local machine 'foofighter' to the remote machine 'tengwar' to runxeyes.Logging intoOpenWrt via SSH usingPuTTY running onWindows.
SSH is a protocol that can be used for many applications across many platforms including mostUnix variants (Linux, theBSDs includingApple'smacOS, andSolaris), as well asMicrosoft Windows. Some of the applications below may require features that are only available or compatible with specific SSH clients or servers. For example, using the SSH protocol to implement aVPN is possible, but presently only with theOpenSSH server and client implementation.
For login to a shell on a remote host (replacingTelnet andrlogin)
For executing a single command on a remote host (replacingrsh)
For setting up automatic (passwordless) login to a remote server (for example, usingOpenSSH[36])
In combination withrsync to back up, copy and mirror files efficiently and securely
The SSH protocol has a layered architecture with three separate components:
Thetransport layer (RFC4253) typically uses theTransmission Control Protocol (TCP) ofTCP/IP, reservingport number 22 as a server listening port. This layer handles initial key exchange as well as server authentication, and sets up encryption, compression, and integrity verification. It exposes to the upper layer an interface for sending and receiving plaintext packets with a size of up to 32,768 bytes each, but more can be allowed by each implementation. The transport layer also arranges for key re-exchange, usually after 1 GB of data has been transferred or after one hour has passed, whichever occurs first.
Theuser authentication layer (RFC4252) handles client authentication, and provides a suite of authentication algorithms. Authentication isclient-driven: when one is prompted for a password, it may be the SSH client prompting, not the server. The server merely responds to the client's authentication requests. Widely used user-authentication methods include the following:
password: a method for straightforward password authentication, including a facility allowing a password to be changed. Not all programs implement this method.
keyboard-interactive (RFC4256): a versatile method where the server sends one or more prompts to enter information and the client displays them and sends back responses keyed in by the user. Used to provideone-time password authentication such asS/Key orSecurID. Used by some OpenSSH configurations whenPAM is the underlying host-authentication provider to effectively provide password authentication, sometimes leading to an inability to log in with a client that supports just the plainpassword authentication method.
GSSAPI authentication methods which provide an extensible scheme to perform SSH authentication using external mechanisms such asKerberos 5 orNTLM, providingsingle sign-on capability to SSH sessions. These methods are usually implemented by commercial SSH implementations for use in organizations, though OpenSSH does have a working GSSAPI implementation.
Theconnection layer (RFC4254) defines the concept of channels, channel requests, and global requests, which define the SSH services provided. A single SSH connection can be multiplexed into multiple logical channels simultaneously, each transferring data bidirectionally. Channel requests are used to relay out-of-band channel-specific data, such as the changed size of a terminal window or the exit code of a server-side process. Additionally, each channel performs its own flow control using the receive window size. The SSH client requests a server-side port to be forwarded using a global request. Standard channel types include:
shell for terminal shells, SFTP and exec requests (including SCP transfers)
direct-tcpip for client-to-server forwarded connections
forwarded-tcpip for server-to-client forwarded connections
TheSSHFP DNS record (RFC 4255) provides the public host key fingerprints in order to aid in verifying the authenticity of the host.
This open architecture provides considerable flexibility, allowing the use of SSH for a variety of purposes beyond a secure shell. The functionality of the transport layer alone is comparable toTransport Layer Security (TLS); the user-authentication layer is highly extensible with custom authentication methods; and the connection layer provides the ability to multiplex many secondary sessions into a single SSH connection, a feature comparable toBEEP and not available in TLS.
In 1998, a vulnerability was described in SSH 1.5 which allowed the unauthorized insertion of content into an encrypted SSH stream due to insufficient data integrity protection fromCRC-32 used in this version of the protocol.[42][43] A fix known as SSH Compensation Attack Detector[44] was introduced into most implementations. Many of these updated implementations contained a newinteger overflow vulnerability[45] that allowed attackers to execute arbitrary code with the privileges of the SSH daemon, typically root.
In January 2001 a vulnerability was discovered that allows attackers to modify the last block of anIDEA-encrypted session.[46] The same month, another vulnerability was discovered that allowed a malicious server to forward a client authentication to another server.[47]
Since SSH-1 has inherent design flaws that make it vulnerable, it is now generally considered obsolete and should be avoided by explicitly disabling fallback to SSH-1. Most modern servers and clients support SSH-2.[47]
In November 2008, a theoretical vulnerability was discovered for all versions of SSH which allowed recovery of up to 32 bits of plaintext from a block of ciphertext that was encrypted using what was then the standard default encryption mode,CBC.[48] The most straightforward solution is to useCTR, counter mode, instead of CBC mode, since this renders SSH resistant to the attack.[48]
On December 28, 2014Der Spiegel published classified information[49] leaked by whistleblowerEdward Snowden which suggests that theNational Security Agency may be able to decrypt some SSH traffic. The technical details associated with such a process were not disclosed. A 2017 analysis of theCIA hacking toolsBothanSpy andGyrfalcon suggested that the SSH protocol was not compromised.[50]
A novel man-in-the-middle attack against most current ssh implementations was discovered in 2023. It was named theTerrapin attack by its discoverers.[51][52] However, the risk is mitigated by the requirement to intercept a genuine ssh session, and that the attack is restricted in its scope, fortuitously resulting mostly in failed connections.[53][54] The ssh developers have stated that the major impact of the attack is to degrade thekeystroke timing obfuscation features of ssh.[54] The vulnerability was fixed in OpenSSH 9.6, but requires both client and server to be upgraded for the fix to be fully effective.
RFC4344 –The Secure Shell (SSH) Transport Layer Encryption Modes
RFC4345 –Improved Arcfour Modes for the Secure Shell (SSH) Transport Layer Protocol
The protocol specifications were later updated by the following publications:
RFC4419 –Diffie-Hellman Group Exchange for the Secure Shell (SSH) Transport Layer Protocol (March 2006)
RFC4432 –RSA Key Exchange for the Secure Shell (SSH) Transport Layer Protocol (March 2006)
RFC4462 –Generic Security Service Application Program Interface (GSS-API) Authentication and Key Exchange for the Secure Shell (SSH) Protocol (May 2006)
RFC4716 –The Secure Shell (SSH) Public Key File Format (November 2006)
RFC4819 –Secure Shell Public Key Subsystem (March 2007)
RFC5647 –AES Galois Counter Mode for the Secure Shell Transport Layer Protocol (August 2009)
RFC5656 –Elliptic Curve Algorithm Integration in the Secure Shell Transport Layer (December 2009)
RFC6187 –X.509v3 Certificates for Secure Shell Authentication (March 2011)
RFC6239 –Suite B Cryptographic Suites for Secure Shell (SSH) (May 2011)
RFC6594 –Use of the SHA-256 Algorithm with RSA, Digital Signature Algorithm (DSA), and Elliptic Curve DSA (ECDSA) in SSHFP Resource Records (April 2012)
RFC6668 –SHA-2 Data Integrity Verification for the Secure Shell (SSH) Transport Layer Protocol (July 2012)
^Seggelmann, R.; Tuxen, M.; Rathgeb, E.P. (18–20 July 2012).SSH over SCTP — Optimizing a multi-channel protocol by adapting it to SCTP. 8th International Symposium on Communication Systems, Networks & Digital Signal Processing (CSNDSP). pp. 1–6.doi:10.1109/CSNDSP.2012.6292659.ISBN978-1-4577-1473-3.S2CID8415240.
^Barrett, Daniel J.; Silverman, Richard E. (2001).SSH, the secure shell: the definitive guide (1st ed.). Cambridge [Mass.]: O'Reilly. p. 11.ISBN978-0-596-00011-0.
^"OSSH Information for VU#419241".CERT Coordination Center. 2006-02-15.Archived from the original on 2007-09-27.Either way ossh is old and obsolete and I don't recommend its use.
^Sobell, Mark (2012).A Practical Guide to Linux Commands, Editors, and Shell Programming (3rd ed.). Upper Saddle River, NJ: Prentice Hall. pp. 702–704.ISBN978-0133085044.
Barrett, Daniel J.; Silverman, Richard E.; Byrnes, Robert G. (2005).SSH: The Secure Shell (The Definitive Guide) (2nd ed.). O'Reilly.ISBN0-596-00895-3.