Case Study: TellusLabs achieves scalable, cost‑efficient, low‑latency predictive global agriculture mapping with ScyllaDB

TellusLabs Taps Scylla to Power Predictive Maps of Global Agriculture

TellusLabs builds a living map of the world’s food supply by turning satellite imagery and meteorological feeds into field‑level time series and predictive crop‑yield models. Faced with 15 years of daily data, 250 metrics per field and a goal of supporting 500 million fields (over 600 trillion records), TellusLabs needed a space‑efficient, low‑latency database that a small team could operate cost‑effectively — so they evaluated and selected ScyllaDB (Scylla) to meet those requirements.

TellusLabs replaced Apache Cassandra with ScyllaDB and deployed a Scylla cluster that delivered 100% better performance while using one‑third the cluster size required by Cassandra, reducing instance count and operational overhead. Migration was fast (the full dataset moved in a few hours), and ScyllaDB’s partitioning also enabled secure multi‑tenant isolation for proprietary customer data — collectively allowing TellusLabs to serve large, real‑time slices of its dataset more cheaply and reliably while scaling to ingest more sensor sources.

TellusLabs

Ophir Horn

Head of Engineering