How It Works
ExyData Historian offloads older, time-keyed rows into partitioned Parquet while keeping them fully queryable.
1
Connect sources
Point Historian at your existing databases. We support:
- InfluxDB (1.x/2.x/3.x)
- Timescale/Postgres
- MySQL
- SQL Server
• Snapshot + tail (CDC) for RDBMS
• Streaming for time-series databases
2
Tier to open storage
We automatically move your old data to cost-effective storage:
- ZSTD-compressed Parquet
Partitioned by date/hour for optimal query performance
- Sidecar partition index
Fast file pruning for efficient queries
- Automated maintenance
Nightly compaction & retention policy enforcement
3
Query in SQL
Access your cold data with familiar SQL tools:
DuckDB HTTP endpoint
Interactive SQL queries with sub-second response times
Optional Trino
Federated queries across multiple data sources
Result: Predicate pushdown + partition pruning = sub-second scans on cold windows
Architecture Highlights
Runs anywhere
On-prem or in your VPC
Open formats
Parquet by default, Iceberg optional
No lock-in
Your bucket, your keys