Federated Identity and Access Management
Galaxy's identity and access management (IAM) is a set of protocols and methods that enable users to login to Galaxy, and securely delegate Galaxy to access their cloud-based resources.
- Jalili, Vahid, et al. "Cloud Bursting Galaxy: Federated Identity and Access Management." bioRxiv (2018): 506238.
Galaxy supports anonymous users, hence enabling researchers to use Galaxy without needing to create a user account. For instance, anyone can use Galaxy Main without having to create a user account. Despite of its convenience, anonymous users can only benefit from a subset of Galaxy's features. For instance, they cannot save their histories, workflows, or visualization. Therefore, to fully appreciate Galaxy's features, researchers are required to have a user account on Galaxy.
For user's convenience, Galaxy offers a wide-variety of methods to create a user account:
- Galaxy Username and Password: This has a built-in mechanism for creating user accounts via a local username and password. Without any additional configuration, Galaxy uses its database to maintain usernames and passwords, and stores passwords encrypted using Password-Based Key Derivation Function 2 (PBKDF2) algorithm.
- OpenID Connect (OIDC): Leveraging the OIDC protocol, login to Galaxy is enabled without explicitly creating a local Galaxy user account. This protocol is the current latest industry-level standard for user authentication, and has been widely adopted by various platforms. This mechanism can be used in combination with the Galaxy username and password option.
- External authentication: This framework allows an instance of Galaxy to delegate authentication to an external authentication system such as Lightweight Directory Access Protocol (LDAP) server, Pluggable Authentication Module (PAM), or an upstream proxy server (e.g., NGINX or Apache). These systems are aconfigured to authenticate users and pass this authentication information along to Galaxy using the HTTP remote user mechanism. Thus, by the time Galaxy is aware of a request, the user identity will have been determined and there will be no need for Galaxy to do any additional authentication work, such as showing a login screen or checking user credentials. However, accepting an identity asserted by the Web server does not relieve Galaxy from having to create a user account for such an identity. Thus, Galaxy automatically creates a user for each identity of this kind, recording that the user is “external” and also creating a random password in order to effectively disable traditionally performed logins for the user.
Depending on your role/interests, please refer to the following pages for more details:
A user can leverage Galaxy to analyze their data stored on a cloud-based storage, and can also store their Galaxy datasets on a cloud-based storage (see this page for details). The ability to store/read data on/from a cloud-based storage, provides users with a (theoretically) unlimited storage space, and facilitates their data sharing through a cloud storage bucket.
To access user's private cloud-based resources, Galaxy requires credentials to sign read/write requests to user's storage account. However, Galaxy does NOT ask for users credentials as there are some drawbacks to that, such as:
user credentials would grant a Galaxy instance with the same level of privileges as the user themselves;
requires manual intervention to be revoked, and can affect all the other services users have setup with their credentials;
it adds liability concerns to Galaxy.
Therefore, for user data security, Galaxy implements the current-latest protocols that allow users to securely delegate Galaxy to access their resources without having to share their credentials. This method leverages the Role-Based Access Control (RBAC) model implemented in cloud-based resource providers, and defines a separate role for a Galaxy instance, whose access can be restricted/revoked independent from user's access. Providing Galaxy with the information about the role, Galaxy will then assume it on-behalf-of the user.
See this page for the details on how we implement this method for different resource providers, and how it can be used.