Architecture

This document covers the underlying design principles of Teleport and a detailed description of Teleport architecture.

Design Principles

Teleport was designed in accordance with the following principles:

  • Off the Shelf Security: Teleport does not re-implement any security primitives and uses well-established, popular implementations of the encryption and network protocols.

  • Open Standards: There is no security through obscurity. Teleport is fully compatible with existing and open standards and other software, including OpenSSH.

  • Cluster-Oriented Design: Teleport is built for managing clusters, not individual servers. In practice this means that hosts and users have cluster memberships. Identity management and authorization happen on a cluster level.

  • Built for Teams: Teleport was created under the assumption of multiple teams operating on several disconnected clusters (production-vs-staging, or perhaps on a cluster-per-customer or cluster-per-application basis).

Core Concepts

The following core concepts are integral to understanding the Teleport architecture.

  • Cluster of Nodes. Unlike traditional SSH service, Teleport operates on a cluster of nodes. A cluster is a set of nodes (servers). There are several ramifications of this:

    • User identities and user permissions are defined and enforced on a cluster level.
    • A node must become a cluster member before any user can connect to it via SSH.
    • SSH access to any cluster node is always performed via a cluster proxy, sometimes called an "SSH bastion".
  • User Account. Unlike traditional SSH, Teleport introduces the concept of a User Account. A User Account is not the same as SSH login. For example, there can be a Teleport user "johndoe" who can be given permission to login as "root" to a specific subset of nodes.

  • Teleport Services. A Teleport cluster consists of three separate services, also called "node roles": proxy, auth and node. Each Teleport node can run any combination of them by passing --role flag to the teleport daemon.

  • User Roles. Unlike traditional SSH, each Teleport user account is assigned a role. Having roles allows Teleport to implement role-based access control (RBAC), i.e. assign users to groups (roles) and restrict each role to a subset of actions on a subset of nodes in a cluster.

  • Certificates. Teleport uses SSH certificates to authenticate nodes and users within a cluster. Teleport does not allow public key or password-based SSH authentication.

  • Dynamic Configuration. Nearly everything in Teleport can be configured via the configuration file, /etc/teleport.yaml by default. But some settings can be changed at runtime, by modifying the cluster state (eg., creating user roles or connecting to trusted clusters). These operations are called 'dynamic configuration'.

User Accounts

Teleport supports two types of user accounts:

  • Local users are created and stored in Teleport's own identitiy storage. A cluster administrator has to create account entries for every Teleport user. Teleport supports second factor authentication (2FA) and it is enforced by default. There are two types of 2FA supported:
  • External users are users stored elsewhere else within an organization. Examples include Github, Active Directory (AD) or any identity store with an OpenID/OAuth2 or SAML endpoint.

Version Warning

SAML, OIDC and Active Directory are only supported in Teleport Enterprise. Please take a look at Teleport Enterprise chapter for more information.

It is possible to have multiple identity sources configured for a Teleport cluster. In this case an identity source (called "connector") will have to be passed to tsh --auth=connector_name login. Local (AKA internal) users connector can be specified via tsh --auth=local login.

Teleport Cluster

Lets explore how these services come together and interact with Teleport clients and with each other.

High Level Diagram of a Teleport cluster

Teleport Overview

Notice that the Teleport Admin tool must be physically present on the same machine where Teleport auth is running. Adding new nodes or inviting new users to the cluster is only possible using this tool.

Once nodes and users (clients) have been invited to the cluster, lets go over the sequence of network calls performed by Teleport components when the client tries to connect to the node.

1) The client tries to establish an SSH connection to a proxy using either the CLI interface or a web browser (via HTTPS). When establishing a connection, the client offers its public key. Clients must always connect through a proxy for two reasons:

  • Individual nodes may not always be reachable from "the outside".
  • Proxies always record SSH sessions and keep track of active user sessions. This makes it possible for an SSH user to see if someone else is connected to a node she is about to work on.

2) The proxy checks if the submitted certificate has been previously signed by the auth server. If there was no key previously offered (first time login) or if the certificate has expired, the proxy denies the connection and asks the client to login interactively using a password and a 2nd factor.

Teleport uses Google Authenticator for the two-step authentication. The password + 2nd factor are submitted to a proxy via HTTPS, therefore it is critical for a secure configuration of Teleport to install a proper HTTPS certificate on a proxy.

Warning

Do not use a self-signed certificate in production! If the credentials are correct, the auth server generates and signs a new certificate and returns it to a client via the proxy. The client stores this key and will use it for subsequent logins. The key will automatically expire after 23 hours by default. This TTL can be configured to a maximum of 30 hours and a minimum of 1 minute.

3) At this step, the proxy tries to locate the requested node in a cluster. There are three lookup mechanisms a proxy uses to find the node's IP address:

  • Tries to resolve the name requested by the client.
  • Asks the auth server if there is a node registered with this nodename.
  • Asks the auth server to find a node (or nodes) with a label that matches the requested name.

If the node is located, the proxy establishes the connection between the client and the requested node and begins recording the session, sending the session history to the auth server to be stored.

4) When the node receives a connection request, it too checks with the auth server to validate the submitted client certificate. The node also requests the auth server to provide a list of OS users (user mappings) for the connecting client, to make sure the client is authorized to use the requested OS login.

In other words, every connection is authenticated twice before being authorized to log in:

  • User's cluster membership is validated when connecting a proxy.
  • User's cluster membership is validated again when connecting to a node.
  • User's node-level permissions are validated before authorizing her to interact with SSH subsystems.

Detailed Diagram of a Teleport cluster

Teleport Everything

Cluster State

Each cluster node is completely stateless and holds no secrets such as keys, passwords, etc. The persistent state of a Teleport cluster is kept by the auth server. There are three types of data stored by the auth server:

  • Cluster State. A Teleport cluster is a set of machines whose public keys are signed by the same certificate authority (CA), with the auth server acting as the CA of a cluster. The auth server stores its own keys in a cluster state storage. Also, all of cluster dynamic configuration is stored there as well, including:

    • Node membership information and online/offline status for each node.
    • List of active sessions.
    • List of locally stored users.
    • RBAC configuration (roles and permissions).
    • Other dynamic configuration.
  • Audit Log. When users log into a Teleport cluster, execute remote commands and logout, that activity is recorded in the audit log. See Audit Log for more details.

  • Recorded Sessions. When Teleport users launch remote shells via tsh ssh command, their interactive sessions are recorded and stored by the auth server. Each recorded session is a file which is saved in /var/lib/teleport, by default but can also be saved in an AWS S3 bucket.

Storage Back-Ends

Different types of cluster data can be configured with different storage back-ends as shown in the table below:

Data Type Supported Back-ends Notes
Cluster state dir, etcd, dynamodb If multi-server (HA) configuration is used with dir backend, users must use NFS or similar networked filesystem to keep the data_dir in sync between the auth servers.
Audit Log Events dir, dynamodb If dynamodb is used for the audit log events, s3 back-end must be used for the recorded sessions.
Recorded Sessions dir, s3 s3 is mandatory if dynamodb is used for the audit log.

Trivia

The reason Teleport designers split the audit log events and the recorded sessions into different back-ends is because of the nature of the data. A recorded session is a compressed binary stream (blob) while the event is a well-defined JSON structure. dir works well enough for both in small deployments, but large clusters require specialized data stores: S3 is perfect for uploading session blobs, while DynamoDB or etcd are better suited to store the cluster state.

The combination of DynamoDB + S3 is especially popular among AWS users because it allows them to run Teleport clusters completely devoid of local state.

Teleport Services

There are three types of services (roles) in a Teleport cluster.

Service (Node Role) Description
node This role provides the SSH access to a node. Typically every machine in a cluster runs this role. It is stateless and lightweight.
proxy The proxy accepts inbound connections from the clients and routes them to the appropriate nodes. The proxy also serves the Web UI.
auth This service provides authentication and authorization service to proxies and nodes. It is the certificate authority (CA) of a cluster and the storage for audit logs. It is the only stateful component of a Teleport cluster.

Although teleport daemon is a single binary, it can provide any combination of these services via --roles command line flag or via the configuration file.

In addition to teleport daemon, there are three client tools you will use:

Tool Description
tctl Cluster administration tool used to invite nodes to a cluster and manage user accounts. tctl must be used on the same machine where auth is running.
tsh Teleport client tool, similar in principle to OpenSSH's ssh. Use it to log into remote SSH nodes, list and search for nodes in a cluster, securely upload/download files, etc. tsh can work in conjunction with ssh by acting as an SSH agent.
Web browser You can use your web browser to log into any Teleport node, just open https://<proxy-host>:3080 (proxy-host is one of the machines that has proxy service enabled).

Let's explore each of the Teleport services in detail.

The Auth Service

The auth server acts as a certificate authority (CA) of the cluster. Teleport security is based on SSH certificates and every certificate must be signed by the cluster auth server.

There are two types of certificates the auth server can sign:

  • Host certificates are used to add new nodes to a cluster.
  • User certificates are used to authenticate users when they try to log into a cluster node.

Upon initialization the auth server generates a public / private keypair and stores it in the configurable key storage. The auth server also keeps the records of what has been happening inside the cluster: it stores recordings of all SSH sessions in the configurable events storage.

Teleport Auth

When a new node joins the cluster, the auth server generates a new public / private keypair for the node and signs its certificate.

To join a cluster for the first time, a node must present a "join token" to the auth server. The token can be static (configured via a config file) or a dynamic, single-use token.

NOTE

When using dynamic tokens, their default time to live (TTL) is 15 minutes, but it can be reduced (not increased) via tctl flag.

Nodes that are members of a Teleport cluster can interact with the auth server using the auth API. The API is implemented as an HTTP REST service running over the SSH tunnel, authenticated using host or user certificates previously signed by the auth server.

All nodes of the cluster send periodic ping messages to the auth server, reporting their IP addresses and values of their assigned labels. The list of connected cluster nodes is accessible to all members of the cluster via the API.

Clients can also connect to the auth API through Teleport proxy to use a limited subset of the API to discover the member nodes of the cluster.

Cluster administration is performed using tctl command line tool.

NOTE

For high availability in production, a Teleport cluster can be serviced by multiple auth servers running in sync. Check HA configuration in the Admin Guide.

The Proxy Service

The proxy is a stateless service which performs two functions in a Teleport cluster:

  1. It serves a Web UI which is used by cluster users to sign up and configure their accounts, explore nodes in a cluster, log into remote nodes, join existing SSH sessions or replay recorded sessions.

  2. It serves as an authentication gateway, asking for user credentials and forwarding them to the auth server via Auth API. When a user executes tsh --proxy=p ssh node command, trying to log into "node", the tsh tool will establish HTTPS connection to the proxy "p" and authenticate before it will be given access to "node".

All user interactions with the Teleport cluster are done via a proxy service. It is recommended to have several of them running.

When you launch the Teleport Proxy for the first time, it will generate a self-signed HTTPS certificate to make it easier to explore Teleport.

Warning

It is absolutely crucial to properly configure TLS for HTTPS when you use Teleport Proxy in production.

Web to SSH Proxy

In this mode, Teleport Proxy implements WSS (secure web sockets) to SSH proxy:

Teleport Proxy Web

  1. User logs in using username, password and 2nd factor token to the proxy.
  2. Proxy passes credentials to the auth server's API
  3. If auth server accepts credentials, it generates a new web session and generates a special ssh keypair associated with this web session. Auth server starts serving OpenSSH ssh-agent protocol to the proxy.
  4. From the SSH node's perspective it's a regular SSH client connection that is authenticated using OpenSSH certificate, so no special logic is needed.

NOTE

Unlike in SSH proxying, in web mode Teleport Proxy terminates the traffic and re-encodes for SSH client connection.

SSH Proxy

Getting signed short lived certificate

Teleport Proxy implements a special method to let clients get short lived certificates signed by auth's host certificate authority:

Teleport Proxy SSH

  1. TSH client or TSH agent generate OpenSSH keypair and forward generated public key and username, password and second factor token that are entered by user to the proxy.
  2. Proxy forwards request to the auth server.
  3. If auth server accepts credentials, it generates a new certificate signed by its user CA and sends it back to the proxy.
  4. Proxy returns the user certificate to the client and client stores it in ~/.tsh/keys

Connecting to the nodes

Once the client has obtained a short lived certificate, it can use it to authenticate with any node in the cluster. Users can use the certificate using standard OpenSSH client (and get it using ssh-agent socket served by tsh agent) or using tsh directly:

Teleport Proxy Web

  1. SSH client connects to proxy and executes proxy subsystem of the proxy's SSH server, providing target node's host and port location.
  2. Proxy dials to the target TCP address and starts forwarding the traffic to the client.
  3. SSH client uses established SSH tunnel to open a new SSH connection and authenticate with the target node using its client certificate.

NOTE

Teleport's proxy command makes it compatible with SSH jump hosts implemented using OpenSSH's ProxyCommand

Certificates

Teleport uses standard SSH certificates for client and host authentication.

One can think of an SSH certificate as a "permit" issued and time-stamped by a certificate authority. A certificate contains four important pieces of data:

  • List of principals (identities) this certificate belongs to.
  • Signature of the certificate authority who issued it, i.e. the auth server.
  • The expiration date, also known as "time to live" or simply TTL.
  • Additional data, often stored as a certificate extension.

One can think of a Teleport auth server as a certificate authority (CA) of a cluster which issues certificates. But reality is a bit more complicated, because cluster hosts need their own certificates too, therefore there are two CAs inside the auth server per cluster:

  • Host CA issues host certificates.
  • User CA issues user certificates.

Host Certificates

All hosts (i.e. nodes, proxies and auth servers) always use host certificates signed by the auth server's host CA to authenticate to join the cluster. Teleport does not allow SSH sessions into nodes that are not cluster members.

A host certificate contains the node's role (like proxy, auth or node) as a certificate extension (opaque signed string). All hosts in a cluster can connect to auth server's HTTP API via SSH tunnel that checks each connecting client's certificate and role to enforce access control (e.g. client connection using node's certificate won't be able to add and delete users, and can only get auth servers registered in the cluster).

User Certificates

The auth server uses its user CA to issue user certificates. In addition to user's identity, user certificates also contain user roles and SSH options, like "permit-agent-forwarding".

This additional data is stored as certificate extensions and is protected by the CA signature.

Certificate Rotation

By default all user certificates have an expiration date, also known as time to live (TTL). This TTL can be configured by a Teleport administrator. But the host certificates issued by an auth server are valid indefinitely by default.

Teleport supports certificate rotation, i.e. the process of invalidating all previously issued certificates regardless of their TTL. Certificate rotation is triggered by tctl auth rotate command. When this command is invoked by a Teleport administrator on one of cluster's auth servers, the following happens:

  • A new certificate authority (CA) key is generated.
  • The old CA will be considered valid alongside the new CA for some period of time. This period of time is called grace period.
  • During the grace period all previously issued certificates will be considered valid, assuming their TTL isn't expired.
  • After the grace period is over, the certificates issued by the old CA are no longer accepted.

This process is repeated twice, for both the host CA and the user CA. Take a look at the admin guide to learn how to use certificate rotation in practice.

Audit Log

The Teleport auth server keeps the audit log of SSH-related events that take place on any node with a Teleport cluster. It is important to understand that the SSH nodes emit audit events and submit them to the auth server.

Compatibility Warning

Because all SSH events like exec or session_start are reported by the Teleport node service, they will not be logged if you are using OpenSSH sshd daemon on your nodes.

Only an SSH server can report what's happening to the Teleport auth server. The audit log is a JSON file which is by default stored on the auth server's filesystem under /var/lib/teleport/log. The format of the file is documented in the Admin Manual.

Teleport users are encouraged to export the events into an external long term storage.

Deployment Considerations

If multiple Teleport auth servers are used to service the same cluster (HA mode) a network file system must be used for /var/lib/teleport/log to allow them to combine all audit events into the same audit log.

Session Recording

By default, destination nodes submit SSH session traffic to the auth server for storage. These recorded sessions can be replayed later via tsh play command or in a web browser.

Some Teleport users mistakenly believe that audit and session recording happen on the Teleport proxy server. This is not the case because a proxy cannot see the encrypted traffic, it is encrypted end-to-end, i.e. from an SSH client to an SSH server/node, see the diagram below:

session-recording-diagram

However, starting from Teleport 2.4 it is now possible to configure the Teleport proxy to enable the "recording proxy mode". In this mode the proxy terminates (decrypts) the SSH connection using the certificate supplied by the client via SSH agent forwarding and then establishes it's own SSH connection to the final destination server, effectively becoming an authorized "man in the middle". This allows the proxy server to forward SSH session data to the auth server to be recorded, as shown below:

recorindg-proxy

The recording proxy mode, although less secure, was added to allow Teleport users to enable session recording for OpenSSH's servers running sshd which is helpful when gradually transitioning large server fleets to Teleport.

We consider the "recording proxy mode" to be less secure for two reasons:

  1. It grants additional privileges to the Teleport proxy. In the default mode the proxy stores no secrets and cannot "see" the decrypted data. This makes a proxy less critical to the security of the overall cluster. But if an attacker gains physical access to a proxy node running in the "recording" mode, they will be able to see the decrypted traffic and client keys stored in proxy's process memory.
  2. Recording proxy mode requires the SSH agent forwarding. Agent forwarding is required because without it a proxy will not be able to establish the 2nd connection to the destination node.

But there are advantages of proxy-based session recording too. When sessions are recorded at the nodes a root user can add iptables rules to prevent sessions logs from reaching the Auth Server. With sessions recorded at the proxy, users with root privileges on nodes have no way of disabling the audit.

See the admin guide to learn how to turn on the recording proxy mode.

Teleport CLI Tools

Teleport offers two command line tools. tsh is a client tool used by the end users, while tctl is used for cluster administration.

TSH

tsh is similar in nature to OpenSSH ssh or scp. In fact, it has subcommands named after them so you can call:

$ tsh --proxy=p ssh -p 1522 [email protected]
$ tsh --proxy=p scp -P example.txt [email protected]/destination/dir

Unlike ssh, tsh is very opinionated about authentication: it always uses auto-expiring keys and it always connects to Teleport nodes via a proxy. It is a mandatory parameter.

When tsh logs in, the auto-expiring key is stored in ~/.tsh and is valid for 23 hours by default, unless you specify another interval via --ttl flag (max of 30 hours and minimum of 1 minute and capped by the server-side configuration).

You can learn more about tsh in the User Manual.

TCTL

tctl is used to administer a Teleport cluster. It connects to the auth server listening on 127.0.0.1 and allows cluster administrator to manage nodes and users in the cluster.

tctl is also a tool which can be used to modify the dynamic configuration of the cluster, like creating new user roles or connecting trusted clusters.

You can learn more about tctl in the Admin Manual.