Integrate, manage, secure, and analyze all of your enterprise data.



Palantir Phoenix is a clusterable data store that supports sub-second queries against trillions of records at petabyte scale. Phoenix leverages several open source technologies to manage data at scale and to perform advanced analytics against those data.


The Palantir Raptor service provides in-place federated searching of external data sources. Raptor utilizes an On-The-Fly (OTF) approach to data integration. When a federated query identifies a Raptor result, that record is transferred to the Revisioning Database on-the-fly.


The Palantir Gotham platform search service provides for full-text querying across all data in the system, both structured and unstructured.


Horizon is an in-memory database created by Palantir to drive interactive workflows on large amounts of data. Horizon allows analysts to query across billions of objects and receive results in about 10 seconds. Originally built in 2009, it's similar in design to Apache Spark™. As the technology that powers Palantir Gotham's Object Explorer, Horizon lets analysts filter large sets of data down to a manageable and interesting subset for more detailed analysis.

Dynamic Ontology

Palantir Gotham's flexibly-defined, object-based data model, the Dynamic Ontology is the means by which data from multiple sources are transformed and integrated from their raw storage formats into data objects and associated properties that represent real objects in the world—people, places, things, events, and the connections between them. Because different organizations conceive of the world in different ways, and because these models change over time, the Dynamic Ontology is defined on a per-case basis, and can be updated as new data sources are added, removed, or simply reconceptualized. This flexible, unified data model greatly simplifies the process of data integration in Palantir Gotham, allowing it to be completed on the order of weeks, as opposed to the multi-year schedule of most enterprise data integration projects.

Revisioning Database

The Revisioning Database, or RevDB, is the Palantir Gotham persistent data store. It powers the platform's access control, auditing, knowledge management, and collaboration capabilities. All data in RevDB is accompanied by a history of its lineage including when it was created or modified, who created or modified it, the data source from which it was derived, and any security or access restrictions associated with the data. This metadata can be directly accessed by the client, providing a context-rich analytic experience for users and enabling secure collaboration between users with different access permissions and/or users pursuing different analytic lines of reasoning. The extensive metadata, security controls, and version control capabilities allow different users to interact with different views of a given object at any given time while maintaining data integrity.


AtlasDB is the data storage container for the Revisioning Database. AtlasDB combines the simplicity and scalability of modern distributed NoSQL data stores with the transactional safety and consistency of traditional SQL databases. AtlasDB layers ACID-compliant transactions on top of any compatible key-value store. Built from the ground-up to be a portable and pluggable transaction API, AtlasDB scales up to data-center scale (or down to laptop scale) with a linear price/performance curve.

Nexus Peering

Palantir Gotham is a distributed system; a Nexus Peered network of Palantir Gotham instances is a distributed system of distributed systems. Each instance of the Palantir Gotham platform maintains its own "nexus" of data in the RevDB. A nexus can incorporate data and analysis from another Palantir Gotham instance through an act of synchronization, or "peering." With the help of vector clocks layered on top of the RevDB, Nexus Peering captures, circulates, and merges changes to data shared across Palantir Gotham instances. Conflicting changes that cannot be resolved automatically are queued up for human review and resolution in a graphical interface. Nexus Peering can account for multiple dynamic ontologies and multiple access control regimes while ensuring that data is always in a consistent state across instances. We developed Nexus Peering to enable users across organizational, functional, and geographic boundaries to securely share and collaboratively analyze data.