This page defines key terms used in TSDB for InfluxDB®.
aggregation
Functions that compute a single value from a group of points—for example, computing the average or sum across a time window.
Related terms: function, selector, transformation
batch
A set of points formatted in line protocol and separated by line feeds (0x0A), sent in a single HTTP request. Batching reduces network overhead and improves HTTP write performance. Aim for 5,000 to 10,000 points per batch, adjusting based on your workload.
Related terms: line protocol, point
continuous query (CQ)
A query that TSDB for InfluxDB® runs automatically on a database at regular intervals. A continuous query (CQ) must include a function in the SELECT clause and a GROUP BY time() clause.
Related term: function
database
A logical container for users, retention policies, continuous queries, and time series data.
Related terms: continuous query, retention policy, user
duration
The attribute of a retention policy that specifies how long data is kept. Data older than the specified duration is deleted automatically.
Related term: retention policy
field
A key-value pair that stores metadata and actual data in TSDB for InfluxDB®. Fields are required in the data structure and are not indexed.
Query tip: Compare fields to tags. Because fields are not indexed, queries that filter on field values must scan all points in the specified time range, which is slower than filtering on tags.
Related terms: field key, field set, field value, tag
field key
The key portion of a field. Field keys are strings that store metadata.
Query tip: Compare field keys to tag keys. Tag keys are indexed; field keys are not.
Related terms: field, field set, field value, tag key
field set
All field key-value pairs in a single point.
Related terms: field, field key, field value, point
field value
The value portion of a field. Field values store actual data and can be strings, floats, integers, or Booleans. Each field value is associated with a timestamp.
Field values are not indexed. Queries that filter on field values scan all points in the specified time range.
TSDB for InfluxDB® indexes tag values, not field values.
Related terms: field, field key, field set, tag value, timestamp
function
A built-in operation in TSDB for InfluxDB® that falls into one of three categories: aggregate functions, selector functions, and transformation functions. For the full list, see General syntax of functions.
Related terms: aggregation, selector, transformation
identifier
A token that names an element in TSDB for InfluxDB®. Identifiers include field keys, tag keys, usernames, and names of continuous queries, databases, measurements, and retention policies.
Related terms: database, field key, measurement, retention policy, tag key, user
line protocol
The text format used to write points to a TSDB for InfluxDB® database.
measurement
A string that describes the data stored in a set of fields, similar to a table name in a SQL database.
metastore
The internal service that stores system state and metadata, including shard metadata and information about users, databases, retention policies, and continuous queries.
Related terms: database, retention policy, user
node
A single server in a TSDB for InfluxDB® instance.
Related term: server
now()
The function that returns the current timestamp of the local server, in nanoseconds.
point
A single data record in TSDB for InfluxDB®, similar to a row in a SQL database table. A point consists of a measurement, a tag set, a field set, and a timestamp. Its identity within a series is determined by its timestamp.
A series cannot contain two points with the same timestamp. If you write a point whose timestamp already exists in the series, the field set for that timestamp becomes the union of the existing and new field sets, and the new values are returned on subsequent queries.
points per second
A deprecated measure of write throughput. Because the TSDB for InfluxDB® data schema supports multiple field values per point, points per second does not accurately reflect write volume. Use values per second instead.
Related terms: point, schema, values per second
query
An operation that retrieves data from a TSDB for InfluxDB® database.
retention policy (RP)
A database-level configuration that defines how long data is stored and the shard group duration. Each database has exactly one retention policy. When you create a database, TSDB for InfluxDB® automatically creates a retention policy named autogen.
A series is uniquely identified by its retention policy, measurement, and tag set.
Related terms: duration, measurement, series, shard duration, tag set
schema
The structure of data in a TSDB for InfluxDB® database, including the database, retention policy, series, measurements, tag keys, tag values, and field keys.
Related terms: database, field key, measurement, retention policy, series, tag key, tag value
selector
Functions that return a single point from a specified range. For the full list, see General syntax of functions.
Related terms: aggregation, function, transformation
series
A collection of points that share the same measurement, tag set, and retention policy.
Field sets do not identify a series.
Related terms: field set, measurement, retention policy, tag set
series cardinality
The total number of unique combinations of databases, measurements, tag sets, and field keys in a TSDB for InfluxDB® instance. High series cardinality increases memory usage and can affect performance.
How cardinality is calculated
Multiply the number of unique values for each tag. For example, a measurement has two tags: email with 3 values and status with 2 values. The series cardinality is 3 × 2 = 6.
| status | |
|---|---|
| alice@example.com | start |
| alice@example.com | finish |
| bob@example.com | start |
| bob@example.com | finish |
| carol@example.com | start |
| carol@example.com | finish |
Dependent tags
A dependent tag is a tag whose values are entirely contained within the value scope of another tag. Dependent tags do not increase series cardinality.
For example, adding a firstname tag to the measurement above does not raise cardinality from 6 to 18 (3 × 2 × 3). Because firstname maps one-to-one with email, it is a dependent tag and cardinality stays at 6.
| status | firstname | |
|---|---|---|
| alice@example.com | start | Alice |
| alice@example.com | finish | Alice |
| bob@example.com | start | Bob |
| bob@example.com | finish | Bob |
| carol@example.com | start | Carol |
| carol@example.com | finish | Carol |
Related terms: field key, measurement, tag key, tag set
server
A physical or virtual machine running a TSDB for InfluxDB® database.
Related term: node
shard
The storage unit that holds the actual data and compressed data for a set of series. Each shard consists of one or more Time-Structured Merge Tree (TSM) files on disk and belongs to exactly one shard group. In a given shard group, all points in a series are stored in the same shard, which means all points in a series are stored in the same TSM file.
Related terms: series, shard duration, shard group, TSM
shard duration
The length of time covered by a single shard group, set by the SHARD DURATION parameter in the retention policy.
For example, with SHARD DURATION set to 1w, each shard group covers one week and holds all points written during that week.
Related terms: database, retention policy, series, shard, shard group
shard group
A logical container that groups shards by time range and retention policy. Every database that contains data must have at least one shard group. All data written within a shard group's time window—its shard duration—is stored in that shard group.
Related terms: database, retention policy, series, shard, shard duration
tag
A key-value pair that stores metadata in TSDB for InfluxDB®. Tags are optional but indexed, which makes tag-based queries faster than field-based queries. Use tags for metadata you filter or group by frequently.
Query tip: Compare tags to fields. Tags are indexed; fields are not.
tag key
The key portion of a tag. Tag keys are indexed strings.
Query tip: Compare tag keys to field keys. Tag keys are indexed; field keys are not.
tag set
All tag key-value pairs in a single point.
tag value
The value portion of a tag. Tag values are indexed strings.
timestamp
The date and time associated with a point, stored in UTC.
For details on how to specify timestamps when writing data, see Overview of data exploration.
Related term: point
transformation
Functions that calculate a value or set of values from a specific set of points without returning an aggregated result. For the full list, see General syntax of functions.
Related terms: aggregation, function, selector
TSM
Time-Structured Merge Tree (TSM) is the storage format used by TSDB for InfluxDB®. TSM provides higher compression and read/write throughput than B+ trees or Log-Structured Merge Trees (LSM trees).
user
TSDB for InfluxDB® has two user types:
Administrator: has read and write access to all databases in an instance, and can manage queries and users.
Standard user: can be granted read-only, write-only, or read/write access to individual databases.
values per second
The recommended metric for measuring write throughput in TSDB for InfluxDB®. Calculate it by multiplying the number of points written per second by the number of fields in each point.
For example, 10 batches per second × 5,000 points per batch × 4 fields per point = 200,000 values per second.
Related terms: batch, field, point, points per second
Write Ahead Log (WAL)
A temporary write cache that buffers incoming points before they are flushed to TSM files on disk. TSDB for InfluxDB® flushes the Write Ahead Log (WAL) when the buffer reaches a size threshold or the data has been cached for too long. This batching reduces how often the system writes to permanent storage.
Points in the WAL are queryable. WAL data persists across restarts—when the TSDB for InfluxDB® process starts, all WAL points are flushed to permanent storage before new writes are accepted.
Related term: TSM