v0.1 vs v0.2
v0.1 supported batch analytics. v0.2 allows for true real-time analytics – as you'd expect from a system like ClickHouse.
The fundamental difference is the write ingestion patterns supported in v0.1 compared to v0.2.
Ingestion Pattern
v0.1: Manual Batch uploads
- Batch writes - Data written to columnstore tables in batches
- Manual optimization - Requires manual compaction and file optimization
- Dual-write pattern - Applications write to OLTP tables, then users have to manually set up batch-load into columnstore
v0.2: streaming inserts/upserts
- zeroETL from Postgres tables - create a columnstore mirror of a Postgres table.
- Streaming ingestion - ingest ~1M transactions per second including inserts, updates, and deletes
- Zero maintenance - Columnstore tables always in a read-optimized state
Feature Comparison
Feature | v0.1 | v0.2 |
---|---|---|
Analytics Performance | ✅ Fast queries | ✅ Fast queries |
Write Pattern | Batch writes only | Streaming + batch writes |
Point Operations | ❌ No updates/deletes | ✅ Full CRUD support |
Object Storage | ✅ Available | 🚧 Coming soon |
Neon Integration | ✅ Available | 🚧 Coming soon |
Table Format | Delta Lake | Iceberg |
Primary Use Case | Data warehousing & archival | Real-time analytics + managed Iceberg |