# Learning Palantir Foundry 🚀 "How do we connect to an on-prem Oracle or SAP in a closed network without poking holes in the firewall?" Data Connection is what clears that first hurdle of enterprise adoption. 📌 Title and Feature URL Title: Data Connection URL: 📝 Overview Data Connection is an application that synchronizes external system data into Foundry for use across the data integration, modeling, and ontology layers. It also supports outbound connections through webhooks and data exports to write data back to external systems. It handles many source types and abstracts away the messy parts — authentication, scheduling, and monitoring — so you can configure pipelines from simple interfaces. 🔧 How It Works Foundry standardizes data connections around three principles: - Robustness: automatic retries, processing in small batches, and integrated health monitoring that warns of failures. Data should be ingested "as-is" from the most primitive source, making Foundry's versioned pipelines the single source of truth for all transformations rather than depending on external preprocessing. - Extensibility: beyond standard integrations (databases, FTPS, HDFS, S3, SFTP), the system accommodates new source types. Because core functions like scheduling and orchestration are standardized, only connection-specific adjustments are needed. - Usability: the system abstracts complexity, letting users configure through simple interfaces instead of manually managing authentication, scheduling, and monitoring. - Key components include agent setup, source configuration, batch/streaming syncs, webhooks, and exports. 🛠 Practical Usage - Access Data Connection from the workspace navigation or the application portal. - Configure a source, then set up a batch or streaming sync to ingest data "as-is." - Concentrate post-ingestion transformations in Foundry pipelines, avoiding preprocessing on the source side. - For write-back, configure outbound integrations using webhooks or exports. 🎯 Use Cases - Connect on-prem Oracle/SAP in a closed network via an agent model (outbound-only) without firewall changes. - Scheduled batch ingestion from diverse sources such as databases, SFTP, and S3. - Outbound integration to write processed results back into external systems. ⚠️ Caveats - The design assumes ingesting data "as-is" and centralizing transforms in Foundry pipelines; transforming on the source side undermines traceability. - Configuring agents and sources requires proper network and authentication setup. - Do not assume external service limits or terms; verify the constraints of each connected source in advance. #PalantirFoundry# #DataIntegration#