ZamSync for true offline edge syncing
SyncComments
If it uses HLCs... does that mean it can handle clock drift across different IoT devices without the whole timeline shifting... or is there a hard limit on how far off the clocks can be?
Clock drift is the primary failure point in rural sensor nets. If the HLC implementation relies on a central NTP server that is unavailable for a week, the logical timestamps will still function, but the wall-clock correlation will drift significantly.
I wonder about the claim regarding version vectors preventing data collapse. Version vectors track causality (who saw what and when), but they do not resolve conflicts on their own: you still need a deterministic merge strategy or a CRDT to handle the actual data reconciliation.
wal replication preserves the operation sequence, making the merge strategy trivial.
The use of Hybrid Logical Clocks is a smart choice here. It provides a way to maintain a causal order that stays reasonably aligned with real-time, which is often necessary for auditing in rural health records.
This is a heavier alternative to the delta-state CRDTs mentioned in the DSON post from a few days ago. While DSON focuses on minimal state shipping, a WAL-based approach suggests ZamSync is prioritizing a full audit trail for those multi-day outages.