man-group / ArcticDB

ArcticDB is a high performance, serverless DataFrame database built for the Python Data Science ecosystem.
http://arcticdb.io
Other
1.23k stars 79 forks source link

Azure AD Authentication support #926

Open phoebusm opened 9 months ago

phoebusm commented 9 months ago

Parent ticket: https://github.com/man-group/ArcticDB/issues/898

Add Azure AD Authentication support Reading material: https://learn.microsoft.com/en-us/python/api/azure-identity/azure.identity.defaultazurecredential?view=azure-python https://github.com/Azure/azure-sdk-for-cpp/blob/main/sdk/identity/azure-identity/README.md#credential-classes

phoebusm commented 9 months ago

Azure AD supports five types of credentials for authentication and it will not be practical to support passing all five types of credentials from user as

  1. Complicated pybining
  2. Much better design of Azure identity library in python than C++
    • Much more work to implement

All five AD credentials request all return the same access token for connection, which has a relatively simple structure:

struct AccessToken final
  {
    /**
     * @brief Token string.
     *
     */
    std::string Token;

    /**
     * @brief A point in time after which the token expires.
     *
     */
    DateTime ExpiresOn;
  };

The suggested design will be, user can specify Token and ExpiresOn on the connection string and a nested struct will be added to the protobuf so these can be passed to the C++ layer.

Concerns:

  1. Users need to keep the expiry of the token in mind
    • If the tokens expire before the end of the operation, the operation will fail
    • Similar to the logic of existing SAS authentication support
  2. Cannot find any documentation regarding AD support on Azurite
    • Will continue the search
    • Last option will be moving the logic to the persistence test (which has the real Azure storage) and user Azure CLI for the setup
phoebusm commented 9 months ago

The development will be put on-hold until the study of other authentication listed on https://github.com/man-group/ArcticDB/issues/898 is done, to maximise code-reuse and provide a uniform and simple API for all authentication