QuestDB Features

Time-series specific features — SAMPLE BY, LATEST ON, temporal JOINs, and more.

SAMPLE BY

Time-series downsampling — aggregate data into time buckets:

// Basic sampling
table.select().sampleBy("1h")

// With FILL strategy
table.select().sampleBy("5m", "PREV")

// Multiple fill strategies (one per selected column)
table.select().sampleBy("1h", ["NULL", "PREV", "LINEAR"])

// Fill with constant value
table.select().sampleBy("1h", [{ constant: 0 }])

// Alignment
table.select().sampleBy("1d", "NULL", "CALENDAR")
table.select().sampleBy("1d", "NULL", "FIRST OBSERVATION")

Fill Strategies

Strategy	Description
`"NONE"`	No fill
`"NULL"`	Fill with NULL
`"PREV"`	Carry forward previous value
`"LINEAR"`	Linear interpolation
`{ constant: v }`	Fill with a constant value

Alignment

Value	Description
`"CALENDAR"`	Align buckets to calendar boundaries
`"FIRST OBSERVATION"`	Align to the first observed timestamp

LATEST ON

Get the latest record per partition key. Automatically uses the designated timestamp column.

// Latest per source
table.select().latestOn("source")

// Multiple partition keys
table.select().latestOn("source", "reading")

Example SQL output:

SELECT * FROM energy_readings LATEST ON ts PARTITION BY source

Temporal JOINs

QuestDB supports time-aware JOINs that match rows by timestamp proximity.

ASOF JOIN

Matches each left row with the closest preceding (or equal) right row by timestamp:

const meterRates = defineTable({
  name: "meter_rates",
  columns: {
    ts: q.timestamp.designated(),
    source: q.symbol(),
    rate_kwh: q.double(),
    cost: q.double(),
  },
  partitionBy: "DAY",
});

const ratesTable = db.table(meterRates);

const rows = await table
  .select()
  .asofJoin(ratesTable, "r", (left, right) =>
    and(left.source.ref(), right.source.ref())
  )
  .execute();
// Result type merges both tables: { ts, source, power_kw, ..., "r.rate_kwh", "r.cost", ... }

LT JOIN

Like ASOF JOIN but matches strictly less than (not equal):

table.select().ltJoin(ratesTable, "r", (left, right) =>
  and(left.source.ref(), right.source.ref()),
  "5s" // optional tolerance
)

SPLICE JOIN

A full ASOF JOIN that includes all records from both sides:

table.select().spliceJoin(ratesTable, "r", (left, right) =>
  and(left.source.ref(), right.source.ref())
)

Standard JOINs

Standard SQL JOINs are also supported:

// INNER JOIN
table.select().innerJoin(other, "o", (left, right) =>
  and(left.id.ref(), right.id.ref())
)

// LEFT JOIN (right columns become nullable)
table.select().leftJoin(other, "o", (left, right) =>
  and(left.id.ref(), right.id.ref())
)

// CROSS JOIN
table.select().crossJoin(other, "o")

Join Result Types

Join results are automatically typed with alias-prefixed column names:

Join Type	Right Columns Nullable?
`asofJoin`	Yes
`ltJoin`	Yes
`spliceJoin`	Yes
`leftJoin`	Yes
`innerJoin`	No
`crossJoin`	No

WHERE timestamp IN

QuestDB's interval shorthand for filtering by timestamp:

// All of 2026
table.select().whereTimestamp("ts", "2026")

// 3-month range starting January 2026
table.select().whereTimestamp("ts", "2026-01;3M")

Generates: WHERE ts IN '2026-01;3M'

Regex Match

QuestDB supports Java regex matching with the ~ operator:

table.select()
  .where((c) => c.source.matches("^solar.*"))

Generates: WHERE source ~ '^solar.*'

QuestDB Features

On this page