# Temporary data storage by application

The Syntho platform processes data securely, within the secure infrastructure of the customer. Below is an overview of how any temporary files are handled.

## **Temporary Files in Use**

1. **Parquet Files:**
   * **Purpose:** Used as staging data prior to writing the processed data to the destination.
   * **Contents:** Contains generated data during intermediate stages of processing.
   * **Lifecycle:**
     * These files are created temporarily and are designed to be deleted upon successful completion, cancelling, or failure of the processing job.
     * In case of an unexpected application failure, these files may remain in the internal storage until the application is restarted. After the restart they will be removed automatically.
   * **Security Controls:**
     * **Access Controls:** The internal storage location is secured with restricted access controls to protect the parquet files during their temporary existence.
     * **Cleanup Mechanisms:** Application-level watchdogs and cleanup routines exist to mitigate the risk of files persisting unnecessarily.
2. **Engine JSON Files:**
   * **Purpose:** Generated to encapsulate job configuration metadata before submission to the Ray cluster for distributed processing.
   * **Contents:** Contains only non-sensitive metadata (e.g., configuration details and application runtime information).
   * **Lifecycle:**
     * These files are typically ephemeral and removed once the job is submitted.
     * If the application crashes at a specific processing point, the engine JSON file with public application data may persist.
   * **Security Consideration:**
     * As the file contains only public and non-sensitive data, it poses no security risk if retained within the secure infra of the customer.