Episode 148 - GluuFederation/identerati-office-hours GitHub Wiki

Title: Yes, we need an identity data lake

• Host: Mike Schwartz, Founder/CEO Gluu
• Guest: Nicholas Bowerman, Solutions Architect, IDMWORKS

Channels

• Linkedin Event
• Youtube Video

Description

If identity is going to power everything from user enablement to holistic security and human processes to agentic workflows, its going to need better data. Better means more accurate, less informal, and incredibly current. To achieve all of this and truly enable identity in an enterprise requires a robust identity data lake that normalizes, formats, and applies policies in a unified way.

Homework

• Achieving Least Privilege: OPA's hidden access risks

• The Identity Security Paradox: When More Tools Create Bigger Blind Spots

• How to begin building a modern IAM architecture

• From Identity Analytics to AI: My Journey into LLMs

Takeaways

• ⚡ Modern IAM solutions lack the ability to normalize and correlate identity data. This has always been true, and it still is for the most part.

• ⚡ Enterprises should evaluate their data management capabilities and to engage in more formal data governance and management capabilities. But who do the tools and best practices actually exist to do this?

• ⚡ Keeping data in sync is logistically difficult--you need to detect changes in sources and effect them in your data lake. But you would also need assured messaging... because what happens if you detect a change, but an error is thrown before it's persisted? Adding more sources just increases the challenge to keep the data accurate.

• ⚡ To make decisions, we not only need data, but also metadata about the data? For example, what is the provenance of the data? Can it be trusted?

• ⚡ Sharing data may trigger important privacy considerations. Copying the data might not be acceptable to the owners of the data. And even if data is copied to a data lake, there could be obligations and restrictions regarding the data.

• ⚡ Perhaps centralizing policy management will provide the justificaiton to organize schema. After all, you can't make policies over what you can't descibe.

Livestream Audio Archive

here