Back to Blog

Engineering

Cloud backupRecoveryData protection

Cloud backups: preserving a moment and planning the recovery

Snapshots, attachment verification, safety copies, and keeping an old device from overwriting a restored account.

Implementation overview of the new cloud backup flow. Availability is controlled by rollout settings; check the app. Snapshots are currently created manually.

ISEA Team

Sync distributes the latest changes. A backup preserves a particular point in time. If you accidentally change a record, sync may promptly distribute that mistake; a backup provides a separate route to an earlier state.

ISEA's new backup flow checks that the archive is complete and handles devices reconnecting after a restore. Old pending edits must not overwrite the records you just recovered.

What a snapshot contains

A snapshot saves the data available on the device making the backup: records, document updates, automation definitions, and their attachments. You can create one without enabling everyday record sync.

If changes from another device have not arrived yet, they will be missing from the snapshot. To include the latest cloud records, let this device finish syncing before you back up.

AI conversations, billing, login credentials, and device preferences are outside this business snapshot. It is neither a complete phone image nor a rewind of every account service.

A photo needs more than a link

A meal record may reference a photo. Saving only a temporary address can leave an unusable image after recovery.

The backup collects both the attachment list and the files. It checks that each file is complete and linked to the right record. Missing local files are downloaded where possible; if a required file is unavailable, backup creation stops and reports an error.

The snapshot is written locally and read back for verification before upload. The server also reads it back and verifies it before marking it available. Retries reuse the same snapshot bytes. Its displayed time remains the original snapshot time, not the time of a later upload retry.

What full recovery changes

Full recovery replaces the corresponding account business data in the cloud. Changes after the snapshot can be overwritten, and records added later can be removed. This requires explicit confirmation.

The initiating device pauses relevant writes and saves a local safety copy. The server also preserves the cloud business state before replacement. If safety-copy creation or validation fails, original business data remains intact while the operation waits for retry or cancellation.

After validation, replacement occurs in one transaction. A failed commit rolls back the replacement instead of leaving a mixture of old and restored records.

Reconnecting an older device

Imagine a second phone has been offline for a week with pending edits. Accepting that old queue immediately after recovery could overwrite the restored data.

ISEA advances the account's recovery generation. Writes carrying the old generation are rejected. A device first preserves exportable old data and queued operations, then clears sync caches and receives the new server state. Editing resumes after it catches up and checks the generation again.

An offline phone is not instantly erased remotely. Protection takes effect when it reconnects and handles the recovery state. Isolated old data remains available for manual inspection rather than being automatically merged back.

Local snapshots and cloud backups

Daily local snapshots and pre-migration safety copies address maintenance on the current device. A cloud backup stores a verified business archive outside that device. Their scope, storage locations, and recovery flows differ.

Cloud snapshots currently need to be created manually. Automatic schedules, storage limits, and retention periods will be explained as the feature rolls out. Enabling sync does not create a historical backup every day.

Continue with how ISEA manages data.