pace
pace copied to clipboard
Data policy IN, dynamic view OUT: PACE is the Policy As Code Engine. It helps you to programatically create and apply a data policy to a processing platform like Databricks, Snowflake or BigQuery (or...
PACE — Policy As Code Engine
The open-source Data Security Engine.
Learn more »
Slack
·
Website
·
Issues
·
Docs
·
Roadmap
·
Book a meeting
PACE is the Policy As Code Engine. It helps you to programmatically create and apply a data policy to a processing platform (like Databricks, Snowflake or BigQuery). Through a data contract, you can apply filters, field transforms, tag-based conditions and access settings to create a view inside a data platform. With Pace, you can enforce policies to data to ensure that data is only used by those allowed and in the way it was intended to be used.
Follow the quickstart if you want to dive right in, join on Slack to discuss with us, and use issues and PR's if you want to contribute or miss a feature!
Motivation and problem
PACE is designed to remove friction and cost from using data in real-world organisational settings. In other words: define and implement a policy to "just build" with data, instead of jumping through hoop after hoop.
If (one of) these sound familiar and you're using one of the currently supported platforms, PACE is worth a try:
- You have to navigate many (competing) policies, constraints and stakeholders to access data.
- The data approval process is complicated, costly and lengthy. "Approval" is only a ticket, configuration is not done automatically.
- Data policies cannot be configured uniformly over hybrid and multi-cloud setups.
- Governance and processing are done in different, unconnected tools.
Positioning
Once installed, PACE sits between your data definitions (often a catalog) and processing platform. The deep dive below provides more background.
Supported platforms
Pace currently supports Collibra, Datahub and Open Data Discovery on the catalog side, connecting to Snowflake, Databricks, Google BigQuery, and PostgreSQL for creating your dynamic views.
Supported policy methods
It's early for PACE (we're in alpha). The following policy methods are currently available, and when put together form "rule sets", the basis of data policies in PACE:
- Field transforms, e.g. "replace everything before the @ in the column
email
, or "nullify thephone number
", including access definitions to differentiate between data consumers. - Filter conditions, e.g. "analysts in Greece can only see transaction from our Greek stores".
- Tag-based rules, e.g. "do
something
if data is taggedGreece
". - Global rule sets, e.g. "All data tagged
PII
should always be masked".
These policy methods can be layered to create a powerful programmatic interface to define, implement, maintain and update policies. Create an issue if you think a valuable policy method is missing!
Let's go
To install and use PACE, you need:
- The PACE app (as Spring Boot app or Docker image)
- The PACE CLI to interact with your deployment
Learn more
Head over to the docs for more info, join on Slack to discuss or reach out to the STRM team for more info and/or to test and implement PACE together.
Deep dive
PACE is built to connect the world of descriptive data tools to the actual data processing platforms (where all that data stuff takes place!).
It's designed to make sure your data governance can follow this pattern:
Various data consumers (1), should only be shown a representation of data (2) that is tailored to who they are and what they're allowed to see (3), regardless of the data catalog (4) in which they explore and find data, and regardless of the data processing platform (5) on which they consume the data.
Definitions
-
Data Consumers
The end-user that uses data for a specific purpose, i.e. a Data Analyst or Data Scientist -
Representation of the data
End-users should not see the raw data, but a representation of these data. This way the access to the raw data can be limited to those that administer these data. -
Tailored to who they are and what they're allowed to see
Classically this is about authentication and authorization. The difference here, is that these are used as inputs for the actual representation of the data, instead of classical RBAC. -
The Data Catalog
Data catalogs are used as a central marketplace for producers and consumers of data. Therefore, a system that is able to create representations of data, should be able to integrate with data catalogs. -
The Data Processing Platform
Data processing platforms often have various responsibilities, but it at least includes a storage layer (e.g. data lake / data warehouse) and an interfacing layer to the storage layer. This is where producers store data and where consumers use data.
Solution
To solve this, PACE focuses on creating representations of data (e.g. by generating views), based on so-called Data Policies.
A Data Policy is a structured (human-defined but machine-readable) document, that aims at capturing how source data should be presented to various principals (i.e. a data accessor), and which transformations should be applied to the data, to create a representation of the source data on the data processing platform.
Data Policies are constructed by retrieving the data schema (the structure of the data) from either a data catalog or a data processing platform. Next, various rule sets can be created, that determine how source data is transformed and/or filtered to create a representation of data on the processing platform. Defining rule sets is a cooperation between various teams: data consumers, data producers, and most important, the legal team.
Want to learn more about how to facilitate this cooperation between various teams? Navigate to https://pace.getstrm.com to see how we can help you!
Brand assets
Looking for a visual of PACE to include somewhere (preferably the blog you're writing about that awesome PACE use case!)? Please use one of the variations of the PACE logo to match your brand style and include the link to this repo: