Data Retention

Premium Feature

Data Retention ensures your event data is never lost during downstream endpoint outages or temporary network disruptions. When a destination service becomes unavailable or returns server-side errors, JENTIS automatically captures and caches incoming event packets in a secure buffer.

Once the connection to the destination is safely re-established, the system automatically replays the cached packets in the order they were received, ensuring complete data delivery to your analytics and marketing stack.

How It Works

1. Automatic Detection: The system continuously monitors response codes and connection stability for all configured vendor destinations.

2. Packet Caching: If a vendor endpoint drops connection, times out, or returns retryable HTTP status codes (e.g., 500, 502, 503, 504, or rate-limit 429), incoming event packets are routed to our isolated storage layer.

3. Queue & Replay: As soon as the vendor endpoint passes health checks, the buffered event packets are replayed sequentially to restore real-time parity.

Important Consideration: Vendor Timestamp Handling

Replaying cached data may cause temporal inconsistencies in your downstream reports, depending on how specific vendor APIs handle incoming event timestamps:

• Event-Time Processing (Recommended): Vendors that ingest and respect the original event timestamp (e.g., standard Google Analytics 4 Measurement Protocol events sent with a time_offset or explicit event timestamp) will correctly place replayed data into its original historical context.

• Received-Time Processing: Vendors that ignore event timestamps and stamp incoming data upon arrival will register replayed packets at the time of delivery rather than the time of user action.

Impact on Reporting: If a vendor processes data using received-time, an outage replayed hours later may result in an artificial spike in activity at the time of replay, alongside a corresponding dip during the original outage window.

Best Practices

• Check your downstream vendors' API documentation to confirm whether they support custom timestamp parameters during server-to-server ingestion.

• When evaluating attribution or performance reports immediately following a known vendor outage, allow time for replayed queues to clear before running analytical queries.