Data & Storage
A database optimized for storing and querying timestamped data points — ideal for metrics, sensor data, financial ticks, and events.
A time-series database is purpose-built for data that arrives in time order and is rarely updated once written. Examples: Prometheus, InfluxDB, TimescaleDB, Amazon Timestream, M3, and Druid. They optimize for high-throughput writes (millions of points per second), efficient storage of densely sampled values (delta encoding, compression), and time-bucketed range queries with aggregations.
Classic time-series workloads include infrastructure metrics (CPU, memory, request rate), IoT and sensor data, financial market data, application performance monitoring, and event streams. The common pattern is "write a lot, append-only, query by time range with downsampling".
Time-series databases typically sacrifice some general-purpose features (joins, complex transactions) to achieve their write throughput and compression. Many deployments tier data — recent data at full resolution, older data downsampled and pushed to cheaper object storage.
Use a time-series database for monitoring, IoT pipelines, financial tick data, and any append-only timestamped workload at scale. A relational database can handle small time-series workloads, but breaks down at high cardinality and ingest rates.
High-cardinality time series (many unique label combinations) can blow up storage and degrade query performance. Time-series databases trade general-purpose querying for ingest and time-range performance.
A storage architecture that manages data as objects (file + metadata + ID) in a flat namespace, optimized for huge amounts of unstructured data.
Splitting a large dataset across multiple machines so that each shard holds a subset of the data and handles a subset of the load.
The number of operations a system can handle per unit of time, often measured in requests per second (RPS) or queries per second (QPS).
A choice between relational databases with strict schemas and ACID guarantees and non-relational databases optimized for scale, flexibility, or specialized workloads.
Atomicity, Consistency, Isolation, Durability — the four properties that traditional database transactions guarantee.
A data structure (typically a B-tree or hash table) that lets a database find rows matching a query without scanning the entire table.