Zum Hauptinhalt springen

Retention of Personal Data

This document describes how the OpenTalk service stores personal data, e.g. to cover GPDR compliance.

What data is stored​

The OpenTalk service collects the following types of data:

  • Personal identification data (name, email address and phone number)
  • Streaming keys
  • Recordings
  • Whiteboard data
  • Votings and protocol responses
  • Logs, depending on the mode used

How is your data collected​

Users directly provide the service with their personal data. Data is collected when a user:

  • Registers on the service
  • Adds streaming information to their account
  • Agrees to be part of a recording
  • Participates in a voting/protocol/whiteboard session

Why is it stored​

Personal data is stored for account management, communication and to enhance user experience.

Object Storage is used to provide access to recordings and whiteboard data.

Where do we store your data​

Personal identification data is stored in a postgres database.

Object Storage (recordings, whiteboard) is stored on a MinIO object storage instance.

Voting and protocol responses are temporarily stored in a Redis database. Entries are automatically deleted after a meeting concludes, which occurs when the last participant leaves the meeting or upon service restarts

How long is it stored​

  • Personal identification data is currently stored indefinitely due to the absence of a purge mechanism. We are actively developing a solution to address this issue
  • Streaming keys are stored indefinitely unless deleted by the user
  • Data in the object storage (e.g. recordings and whiteboard data) is stored until the associated event is deleted via a cleanup job or manual deletion
  • Logs don't contain any personal data by default. Altering the log level to a more verbose variant beyond INFO, which we dont recommend, can lead to the inclusion of personal data in the logs

Cleanup Jobs​

These jobs ensure the deletion of data associated with their event types, with the threshold for deletion being configurable via a parameter. Currently, these jobs are not executed automatically and require manual setup by the administrator.