What is QuestDB?

Updated

QuestDB is an open-source time-series database built for demanding workloads. It combines ultra-low latency, high throughput, tiered storage, and fast SQL on open formats, so you can ingest millions of events per second, compute on live data in milliseconds, and keep years of history queryable through one SQL engine.

What QuestDB does

QuestDB is designed around a single idea: the live stream and the complete record should be unified. Instead of stitching together a message queue, a stream processor, and a separate historical store, QuestDB spans messaging, stream processing, and tiered storage in one system.

Its stated capabilities include:

  • High-throughput ingestion — millions of events per second
  • Low-latency computation — results on live data in milliseconds
  • Long retention — years of history remain queryable
  • One SQL engine — standard SQL across live and historical data
  • Open formats — data accessible through Apache Parquet, Iceberg, and Arrow

Published benchmark figures on QuestDB v10.0 cite 19M rows/sec/server ingestion, 220M rows/sec streamed to Arrow, 2.3s to stream 500M rows, and 32ms for the first Arrow batch.

Who it is for

QuestDB targets workloads where dropping a single event is not acceptable. Its site names three main areas:

Domain Typical use
Capital markets Market data and trades, pre-trade signals, research, TCA, risk, P&L, surveillance, compliance
Aerospace & robotics High-volume telemetry and event capture
Energy & infrastructure Continuous operational data at scale

The capital markets section describes a three-step pattern: capture every event once (ticks, orders, executions, transactions, with durability, deduplication, and out-of-order correction built in), act before the moment passes (intraday risk and P&L, surveillance conditions, published results while the market moves), and investigate with the full record (TCA, markouts, post-trade analysis, compliance, and historical research in SQL).

Why the open-format angle matters

Because QuestDB exposes the same data through Parquet, Iceberg, and Arrow, teams that prefer different tools can still work from one record. A quant team can query in SQL while a research or analytics team reads the same data in Arrow or Parquet — no separate copy and no reconciliation step.

How to evaluate it

QuestDB offers a free Enterprise trial and a live demo, so you can test ingestion rates and query latency against your own data rather than relying on published numbers. The site also publishes customer stories, including one describing 7,000+ queries/second at national scale for 90M users, and quotes from users in trading analytics and transaction throughput.

If your problem is high-volume, time-stamped events that need both immediate action and long-term SQL analysis, QuestDB is built for exactly that shape of workload. If your data is low-volume or not time-ordered, a general-purpose database is likely a simpler fit.