Limits

The sizes and counts KDB allows, what happens at each limit, and the features that are not there yet.

• Category: Limits
• Keywords: limit, limits, maximum, max, how big, how many, too big, too many, nested too deeply, 16 MB, 64 MB, 10 million, DECIMAL scale, name length, too many open cursors, lock timeout, database is busy, table is busy, deadlock, idle transaction, request body, 10,000 rows, truncated, disk full, 507, 53100, statement timeout, max-queries-per-user, max-groups, max-result-memory, max_spill_bytes, KDB_NO_ZONE_MAPS, KDB_NO_COLUMN_MASK, KDB_NO_UNION_PUSHDOWN, switches, reader yield, not supported, missing features, known limitations, not available
• See also: Performance tips, Cursors (limits), Transactions, Run SQL over HTTP (/exec, /exp), How KDB stores data, Protection - timeouts, limits, safe updates, read-only

These are the limits of KDB 0.2.0-dev (September 2026). Most of them are there to protect memory and to give a clear error instead of a crash. When a limit is reached, the statement fails with an error that names the limit, and nothing of that statement is saved.

The full list of missing features is in known limitations. Please check again after you update KDB.

Names

Syntax:

database names   1 to 64 characters: letters, digits, _ and -
user names       1 to 63 characters: letters, digits, _ and -
passwords        12 characters to 1024 bytes

Example: A database name that is too long (shows an error)

CREATE DATABASE stock_market_data_from_2015_to_2026_all_symbols_minute_candles_copy;

Result:

Error: line 1, column 17: bad database name 'stock_market_data_from_2015_to_2026_all_symbols_minute_candles_copy': the name is longer than 64 characters

Values

Syntax:

DECIMAL(scale)       scale 0 to 18, 18 digits
NUMERIC(p, scale)    up to 38 digits
STRING, JSON, BYTES  at most 16 MB per value
TIMESTAMP            milliseconds, UTC

Example: DECIMAL scale (shows an error)

SELECT CAST(12.5 AS DECIMAL(19));

Result:

Error: line 1, column 29: expected a scale from 0 to 18 in DECIMAL(scale) but found '19'

Example: A value too big for INT (sample data: sample; shows an error)

INSERT INTO brokers VALUES (3000000000, 'Too Big Securities');

Result:

Error: column 'id' (INT): 3000000000 is out of range (-2147483647 to 2147483647)

Example: A text result over 16 MB (shows an error)

SELECT length(repeat('x', 20000000));

Result:

Error: line 1, column 15: repeat: the result would be larger than 16 MB

Nesting

Syntax:

expressions                  about 100 levels of ( ... )
subqueries and views         16 levels
saved functions, CALL, triggers   16 levels each
tables in one query          63

Example: Too many brackets (shows an error)

SELECT (((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((1)))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))))));

Result:

Error: line 1, column 107: expression is nested too deeply; expected at most 100 levels of nesting

Rows

Syntax:

kdb sql / kdb shell     1,000 rows per result in table format (--max-rows, .maxrows)
HTTP /exec              10,000 rows by default, at most 1,000,000 (limit=)
HTTP /exp               1,000,000 rows by default, any number with limit= (streamed)
one transaction         10 million changed rows
one query (kdb serve)   1 GB of rows in memory, 5,000,000 groups

Command line: kdb sql shows at most --max-rows rows (sample data: sample)

kdb sql market "SELECT time, symbol, close FROM ochl_minutes" --max-rows 2

Output:

time                     | symbol | close
-------------------------+--------+------
2026-09-09T05:15:00.000Z | CITY   | 512.1
2026-09-09T05:16:00.000Z | CITY   | 511.9
(2 rows, more not shown, 0.3 ms)

HTTP: HTTP: limit=2, and truncated is true (sample data: sample)

GET /exec?query=SELECT%20time,%20symbol,%20close%20FROM%20ochl_minutes&limit=2

Response:

HTTP 200 (application/json)

{
  "query": "SELECT time, symbol, close FROM ochl_minutes",
  "results": [
    {
      "statement": "SELECT time, symbol, close FROM ochl_minutes",
      "columns": [
        {"name": "time", "type": "TIMESTAMP"},
        {"name": "symbol", "type": "SYMBOL"},
        {"name": "close", "type": "DOUBLE"}
      ],
      "dataset": [
        ["2026-09-09T05:15:00.000Z", "CITY", 512.1],
        ["2026-09-09T05:16:00.000Z", "CITY", 511.9]
      ],
      "count": 2,
      "truncated": true,
      "rows_affected": 0,
      "message": "",
      "timings": {"parse_us": 30, "execute_us": 287}
    }
  ],
  "columns": [
    {"name": "time", "type": "TIMESTAMP"},
    {"name": "symbol", "type": "SYMBOL"},
    {"name": "close", "type": "DOUBLE"}
  ],
  "dataset": [
    ["2026-09-09T05:15:00.000Z", "CITY", 512.1],
    ["2026-09-09T05:16:00.000Z", "CITY", 511.9]
  ],
  "count": 2,
  "timings": {"parse_us": 30, "execute_us": 287}
}

Writers, readers and transactions

Syntax:

writers                 one per table; writers of different tables run at the same time
waiting for a table     5 seconds unless set (--lock-timeout, SET lock_timeout), then "table 'x' is busy"
idle transaction        rolled back after 30 seconds without a statement (--idle-in-transaction-timeout, SET idle_in_transaction_session_timeout)
open cursors            16 per database, 10 million rows each, closed after 10 unused minutes

See cursors and transactions.

Example: The 17th open cursor (shows an error)

DECLARE c1 CURSOR FOR SELECT 1 AS x;
DECLARE c2 CURSOR FOR SELECT 2 AS x;
DECLARE c3 CURSOR FOR SELECT 3 AS x;
DECLARE c4 CURSOR FOR SELECT 4 AS x;
DECLARE c5 CURSOR FOR SELECT 5 AS x;
DECLARE c6 CURSOR FOR SELECT 6 AS x;
DECLARE c7 CURSOR FOR SELECT 7 AS x;
DECLARE c8 CURSOR FOR SELECT 8 AS x;
DECLARE c9 CURSOR FOR SELECT 9 AS x;
DECLARE c10 CURSOR FOR SELECT 10 AS x;
DECLARE c11 CURSOR FOR SELECT 11 AS x;
DECLARE c12 CURSOR FOR SELECT 12 AS x;
DECLARE c13 CURSOR FOR SELECT 13 AS x;
DECLARE c14 CURSOR FOR SELECT 14 AS x;
DECLARE c15 CURSOR FOR SELECT 15 AS x;
DECLARE c16 CURSOR FOR SELECT 16 AS x;
DECLARE c17 CURSOR FOR SELECT 17 AS x;

Result:

DECLARE c1 CURSOR FOR SELECT 1 AS x

cursor 'c1' declared: 1 row(s)

DECLARE c2 CURSOR FOR SELECT 2 AS x

cursor 'c2' declared: 1 row(s)

DECLARE c3 CURSOR FOR SELECT 3 AS x

cursor 'c3' declared: 1 row(s)

DECLARE c4 CURSOR FOR SELECT 4 AS x

cursor 'c4' declared: 1 row(s)

DECLARE c5 CURSOR FOR SELECT 5 AS x

cursor 'c5' declared: 1 row(s)

DECLARE c6 CURSOR FOR SELECT 6 AS x

cursor 'c6' declared: 1 row(s)

DECLARE c7 CURSOR FOR SELECT 7 AS x

cursor 'c7' declared: 1 row(s)

DECLARE c8 CURSOR FOR SELECT 8 AS x

cursor 'c8' declared: 1 row(s)

DECLARE c9 CURSOR FOR SELECT 9 AS x

cursor 'c9' declared: 1 row(s)

DECLARE c10 CURSOR FOR SELECT 10 AS x

cursor 'c10' declared: 1 row(s)

DECLARE c11 CURSOR FOR SELECT 11 AS x

cursor 'c11' declared: 1 row(s)

DECLARE c12 CURSOR FOR SELECT 12 AS x

cursor 'c12' declared: 1 row(s)

DECLARE c13 CURSOR FOR SELECT 13 AS x

cursor 'c13' declared: 1 row(s)

DECLARE c14 CURSOR FOR SELECT 14 AS x

cursor 'c14' declared: 1 row(s)

DECLARE c15 CURSOR FOR SELECT 15 AS x

cursor 'c15' declared: 1 row(s)

DECLARE c16 CURSOR FOR SELECT 16 AS x

cursor 'c16' declared: 1 row(s)

Error: too many open cursors (16); CLOSE the ones you do not need (an unused cursor also closes after 10 minutes)

Example: DDL inside a transaction is refused (shows an error)

BEGIN;
CREATE TABLE notes (id LONG PRIMARY KEY, body STRING);

Result:

transaction started

Error: CREATE TABLE cannot be used inside a transaction; COMMIT or ROLLBACK first (the transaction was rolled back)

Server and HTTP

Syntax:

request body     64 MB (also after gzip or deflate is unpacked)
request headers  64 KB
idle connection  closed after 30 seconds
worker threads   one per CPU thread, at least 8 (--http-threads)
pg connections   500 (--max-connections)

See server security and HTTP exec.

Command line: The server's defaults are in its banner (sample data: sample)

kdb serve ./market

Output:

KDB 0.2.0-dev
  console   http://127.0.0.1:9900
  data      ./market
  backups   ./_backups/market (from default)
  access    no token
  capacity  32 HTTP threads (default), 500 PostgreSQL connections (default), lock wait 5 s (default)
  lanes     reads using 100 ms of CPU (three times in a row) wait for one of 16 heavy slots, at most 16 HTTP in line (--lanes on)
  reads     32 reads run at once (the CPUs, at least 2; follows the CPU limit), the rest wait in line (--read-slots)
  commits   synced to disk before they are confirmed (--sync-commits always)
  time zone Asia/Katmandu (the computer's); also timers and backup schedules
  readers   read while a long trigger or CALL writes (--reader-yield on); idle open transaction rolled back after 30 s (default)
  memory    26.95 GB for the whole process (85% of the machine's memory); table data 512.0 MB of it (--memory)
  cpu       32 CPUs (the machine's); 16 query threads (half of the CPUs, follows the limit), limits checked every 10 s
  limits    statement timeout 1 min, result memory 1.00 GB, 5000000 groups, 32 statements per user
  safety    writes allowed; DROP TABLE kept 7 days (UNDROP TABLE); audit log off
  logs      slow queries >= 1000 ms to ./market/_logs/slow_queries.jsonl; pg_stat_statements on; metrics at /metrics
  recovery  manifest copy: a, last shutdown: clean
  open      opened in 2 ms: manifest 1 ms, 1 files read 0 ms (8 table files left for their first use), log check 1 ms, replay 0 ms (0 transactions), clean-up 0 ms

Press Ctrl+C to stop.

Disk and memory

Syntax:

free disk space     writes stop below 64 MB free
disk full           HTTP 507 {"code":"disk full"}, PostgreSQL SQLSTATE 53100
old file versions   2 per file, then _trash for 1 hour
log archive         1 GB per database by default; SET WAL_ARCHIVE changes it
memory for data     --memory, default 512 MB for all open databases together

Command line: Sizes and log positions of the files (sample data: sample)

kdb info market

Output:

database        market
manifest        copy a, sequence 9
last shutdown   clean
log position    30 (saved in files up to 30)
newest change   2026-10-22 11:28:27
features        batched_log_sync, block_stats, compressed_log, partition_format_2
manifest keys   26

FILE                                      VERSION       SIZE          LSN  STATUS
catalog                                         6     2.1 KB           30  ok
t/1/all                                         7      777 B           30  ok
t/2/all                                         8      556 B           30  ok
t/3/2025                                        9     1.7 KB           30  ok
t/3/2026                                       10     3.6 KB           30  ok
t/4/2026-09                                    11    57.0 KB           30  ok
t/5/2026-09-09                                 12     1.4 KB           30  ok
t/5/2026-09-10                                 13     2.4 KB           30  ok
t/6/2026-09-10                                 14     2.7 KB           30  ok

9 files, 72.1 KB, 0 marked damaged
(STATUS is what the manifest records. To read every block and find new damage, run: kdb check)

Switches for measuring

These environment variables turn one speed-up off, so you can time a query with and without it. The answers are the same either way. Set them for kdb serve only while you measure.

For one query, /exec?cache=0 skips the result cache (see performance tips).

Not supported yet

Syntax:

NATURAL JOIN   DDL in a transaction

Short list. The details, and the ways around each one, are in known limitations.

Example: NATURAL JOIN is not there yet (sample data: sample; shows an error)

SELECT * FROM brokers b NATURAL JOIN brokers c;

Result:

Error: line 1, column 25: expected JOIN, LEFT JOIN, RIGHT JOIN, FULL JOIN, ASOF JOIN, LT JOIN, SPLICE JOIN, CROSS JOIN, WHERE, LATEST ON, SAMPLE BY, GROUP BY, UNION ALL, ORDER BY, LIMIT, ';' or the end of the input but found 'NATURAL'