Can Edge Computing function without an active internet connection to the central cloud?

All answers to this question.

This is known as "Offline-First" or "Disconnected Edge" architecture. Most modern edge platforms, like AWS IoT Greengrass or Azure IoT Edge, allow you to deploy Lambda functions or containers that run locally regardless of cloud status. For data, you must implement a "Store and Forward" mechanism with a strictly defined TTL (Time To Live). When the local storage hits a threshold—say 80%—the system should start purging non-critical logs or using "Lossy Compression" to save space for high-priority alerts. Once the satellite link is restored, the edge node uses a delta-sync to only upload the most recent and critical data packets.

   Answered 2024-07-20 by Melissa Adams

Regarding the purging of data, how do you programmatically determine which data is "non-critical" during a long-term outage without risking the loss of vital forensic information?

   Answered 2024-08-02 by David Wright

• David, we use "Intelligent Tiering" on the local disk. High-frequency sensor "noise" is averaged into minute-by-minute summaries after 24 hours, which saves about 90% of the space. However, "Event Triggers"—like a sudden pressure spike—are flagged as "Protected" and are never deleted until an acknowledgment signal is received from the central server. This way, we maintain a high-resolution history of anomalies while sacrificing the high-resolution "normal" data that doesn't provide any new insights to the maintenance teams.

     Commented 2024-08-10 by Brian Foster

Kubernetes at the edge is tricky here. You need a distribution that doesn't "panic" and restart all your pods just because it can't reach the central API control plane.

   Answered 2024-08-25 by Kevin Smith

• Exactly, Kevin. Tools like KubeEdge are designed precisely for this, allowing nodes to remain operational even when the master node in the cloud is unreachable.

     Commented 2024-09-01 by Melissa Adams