- Datanet's highly ambitious goal is to provide Ubiquotous Write Thru Caching:
- - CRDT algorithms make modifications locally & then send Deltas system-wide
- - System-wide Deltas are applied in arbitrary ordering & strong consistency is still guaranteed
- - Local write lhrough caches are placed anywhere in the stack & all caches achieve consistency
Agents modify locally cached data and immediately return to computation. Replication happens lazily outside the fast path in an asynchronous flow. Cache-read-hits and writes being done locally results in a massive improvement in processing speed. Lazy replication is both more efficient and increases system robustness. Finally, CRDT algorithms mean strong data guarantees apply.
When an Agent modifies data a Delta is produced. This Delta is then replicated system-wide where other Agent's caching this data apply the Delta and achieve consistency. CRDT algorithms automatically resolve conflicts that happen when two Agents concurrently modify the same data and versioning systems insure causal consistency of Delta applications. Place Datanet Agents anywhere in the stack you want a globally-replicated strongly-consistent highly-robust writable data-cache.
Datanet's long term highly ambitious goal is to provide Ubiquitous Write Though Caches. Place write though caches in application servers, highly available stateful application server clusters, multi-data-center consistent memcache-clusters, browsers, mobile-phones, IoT devices, at the content delivery network edge, etc... These caches all share the same data space and are all made globally consistent with optimal latency. Hopefully the list of quasi-futurustic use-cases will grow ever longer.