Amazon Redshift SQL translation guide

This document details the similarities and differences in SQL syntax between Amazon Redshift and BigQuery to help you plan your migration. Use batch SQL translation to migrate your SQL scripts in bulk, or interactive SQL translation to translate ad hoc queries.

The intended audience for this guide is enterprise architects, database administrators, application developers, and IT security specialists. It assumes you are familiar with Amazon Redshift.

Data types

This section shows equivalents between data types in Amazon Redshift and in BigQuery.

Amazon Redshift BigQuery Notes
Data type Alias Data type
SMALLINT INT2 INT64 Amazon Redshift's SMALLINT is 2 bytes, whereas BigQuery's INT64 is 8 bytes.
INTEGER

INT, INT4

 INT64 Amazon Redshift's INTEGER is 4 bytes, whereas BigQuery's INT64 is 8 bytes.
BIGINT INT8 INT64 Both Amazon Redshift's BIGINT and BigQuery's INT64 are 8 bytes.
DECIMAL NUMERIC NUMERIC
REAL FLOAT4 FLOAT64 Amazon Redshift's REAL is 4 bytes, whereas BigQuery's FLOAT64 is 8 bytes.
DOUBLE PRECISION

FLOAT8, FLOAT

 FLOAT64
BOOLEAN BOOL BOOL Amazon Redshift's BOOLEAN can use TRUE, t, true, y, yes, and 1 as valid literal values for true. BigQuery's BOOL data type uses case-insensitive TRUE.
CHAR

CHARACTER, NCHAR, BPCHAR

STRING
VARCHAR

CHARACTER VARYING, NVARCHAR, TEXT

STRING
DATE DATE
TIMESTAMP TIMESTAMP WITHOUT TIME ZONE DATETIME
TIMESTAMPTZ

TIMESTAMP WITH TIME ZONE

TIMESTAMP Note: In BigQuery, time zones are used when parsing timestamps or formatting timestamps for display. A string-formatted timestamp might include a time zone, but when BigQuery parses the string, it stores the timestamp in the equivalent UTC time. When a time zone is not explicitly specified, the default time zone, UTC, is used. Time zone names or offset from UTC using (-|+)HH:MM are supported, but time zone abbreviations such as PDT are not supported.
GEOMETRY GEOGRAPHY Support for querying geospatial data.

BigQuery also has the following data types that do not have a direct Amazon Redshift analog:

Implicit conversion types

When migrating to BigQuery, you need to convert most of your Amazon Redshift implicit conversions to BigQuery's explicit conversions except for the following data types, which BigQuery implicitly converts.

BigQuery performs implicit conversions for the following data types:

From BigQuery type To BigQuery type

INT64

FLOAT64

INT64

NUMERIC

NUMERIC

FLOAT64

BigQuery also performs implicit conversions for the following literals:

From BigQuery type To BigQuery type
STRING literal
(e.g. "2008-12-25")

DATE

STRING literal
(e.g. "2008-12-25 15:30:00")

TIMESTAMP

STRING literal
(e.g. "2008-12-25T07:30:00")

DATETIME

STRING literal
(e.g. "15:30:00")

TIME

Explicit conversion types

You can convert Amazon Redshift data types that BigQuery doesn't implicitly convert using BigQuery's CAST(expression AS type) function or any of the DATE and TIMESTAMP conversion functions.

When migrating your queries, change any occurrences of the Amazon Redshift CONVERT(type, expression) function (or the :: syntax) to BigQuery's CAST(expression AS type) function, as shown in the table in the Data type formatting functions section.

Query syntax

This section addresses differences in query syntax between Amazon Redshift and BigQuery.

SELECT statement

Most Amazon Redshift SELECT statements are compatible with BigQuery. The following table contains a list of minor differences.

Amazon Redshift BigQuery

SELECT TOP number expression
FROM table

SELECT expression
FROM table
ORDER BY expression DESC
LIMIT number

SELECT
x/total AS probability,
ROUND(100 * probability, 1) AS pct
FROM raw_data

Note: Redshift supports creating and referencing an alias in the same SELECT statement.

SELECT
x/total AS probability,
ROUND(100 * (x/total), 1) AS pct
FROM raw_data

BigQuery also supports the following expressions in SELECT statements, which do not have a Amazon Redshift equivalent:

FROM clause

A FROM clause in a query lists the table references that data is selected from. In Amazon Redshift, possible table references include tables, views, and subqueries. All of these table references are supported in BigQuery.

BigQuery tables can be referenced in the FROM clause using the following:

  • [project_id].[dataset_id].[table_name]
  • [dataset_id].[table_name]
  • [table_name]

BigQuery also supports additional table references:

JOIN types

Both Amazon Redshift and BigQuery support the following types of join:

  • [INNER] JOIN
  • LEFT [OUTER] JOIN
  • RIGHT [OUTER] JOIN
  • FULL [OUTER] JOIN
  • CROSS JOIN and the equivalent implicit comma cross join.

The following table contains a list of minor differences.

Amazon Redshift BigQuery

SELECT col
FROM table1
NATURAL INNER JOIN
table2

SELECT col1
FROM table1
INNER JOIN
table2
USING (col1, col2 [, ...])


Note: In BigQuery, JOIN clauses require a JOIN condition unless the clause is a CROSS JOIN or one of the joined tables is a field within a data type or an array.

WITH clause

A BigQuery WITH clause contains one or more named subqueries that execute when a subsequent SELECT statement references them. Amazon Redshift WITH clauses behave the same as BigQuery's with the exception that you can evaluate the clause once and reuse its results.

Set operators

There are some minor differences between Amazon Redshift set operators and BigQuery set operators. However, all set operations that are feasible in Amazon Redshift are replicable in BigQuery.

Amazon Redshift BigQuery

SELECT * FROM table1
UNION
SELECT * FROM table2

SELECT * FROM table1
UNION DISTINCT
SELECT * FROM table2

Note: Both BigQuery and Amazon Redshift support the UNION ALL operator.

SELECT * FROM table1
INTERSECT
SELECT * FROM table2

SELECT * FROM table1
INTERSECT DISTINCT
SELECT * FROM table2

SELECT * FROM table1
EXCEPT
SELECT * FROM table2

SELECT * FROM table1
EXCEPT DISTINCT
SELECT * FROM table2

SELECT * FROM table1
MINUS
SELECT * FROM table2

SELECT * FROM table1
EXCEPT DISTINCT
SELECT * FROM table2

SELECT * FROM table1
UNION
SELECT * FROM table2
EXCEPT
SELECT * FROM table3

SELECT * FROM table1
UNION ALL
(
SELECT * FROM table2
EXCEPT
SELECT * FROM table3
)


Note: BigQuery requires parentheses to separate different set operations. If the same set operator is repeated, parentheses are not necessary.

ORDER BY clause

There are some minor differences between Amazon Redshift ORDER BY clauses and BigQuery ORDER BY clauses.

Amazon Redshift BigQuery
In Amazon Redshift, NULLs are ranked last by default (ascending order). In BigQuery, NULLs are ranked first by default (ascending order).

SELECT *
FROM table
ORDER BY expression
LIMIT ALL

SELECT *
FROM table
ORDER BY expression



Note: BigQuery does not use the LIMIT ALL syntax, but ORDER BY sorts all rows by default, resulting in the same behavior as Amazon Redshift's LIMIT ALL clause. We highly recommend including a LIMIT clause with every ORDER BY clause. Ordering all result rows unnecessarily degrades query execution performance.

SELECT *
FROM table
ORDER BY expression
OFFSET 10

SELECT *
FROM table
ORDER BY expression
LIMIT count OFFSET 10



Note: In BigQuery, OFFSET must be used together with a LIMIT count. Make sure to set the count INT64 value to the minimum necessary ordered rows. Ordering all result rows
unnecessarily degrades query execution performance.

Conditions

The following table shows Amazon Redshift conditions, or predicates, that are specific to Amazon Redshift and must be converted to their BigQuery equivalent.

Amazon Redshift BigQuery

a = ANY (subquery)

a = SOME (subquery)

a IN subquery

a <> ALL (subquery)

a != ALL (subquery)

a NOT IN subquery

a IS UNKNOWN

expression ILIKE pattern

a IS NULL

LOWER(expression) LIKE LOWER(pattern)

expression LIKE pattern ESCAPE 'escape_char'

expression LIKE pattern


Note: BigQuery does not support custom escape characters. You must use two backslashes \\ as escape characters for BigQuery.

expression [NOT] SIMILAR TO pattern

IF(
LENGTH(
REGEXP_REPLACE(
expression,
pattern,
''
) = 0,
True,
False
)


Note: If NOT is specified, wrap the above IF expression in a NOT expression as shown below:

NOT(
IF(
LENGTH(...
)

expression [!] ~ pattern

[NOT] REGEXP_CONTAINS(
expression,
regex
)

Functions

The following sections list Amazon Redshift functions and their BigQuery equivalents.

Aggregate functions

The following table shows mappings between common Amazon Redshift aggregate, aggregate analytic, and approximate aggregate functions with their BigQuery equivalents.

Amazon Redshift BigQuery
APPROXIMATE COUNT(DISTINCT expression) APPROX_COUNT_DISTINCT(expression)
APPROXIMATE PERCENTILE_DISC(
percentile
) WITHIN GROUP (ORDER BY expression)		 APPROX_QUANTILES(expression, 100)
[OFFSET(CAST(TRUNC(percentile * 100) as INT64))]
AVG([DISTINCT] expression) AVG([DISTINCT] expression)
COUNT(expression) COUNT(expression)
LISTAGG(
[DISTINCT] aggregate_expression
[, delimiter] )
[WITHIN GROUP (ORDER BY order_list)] 		STRING_AGG(
[DISTINCT] aggregate_expression
[, delimiter]
[ORDER BY order_list] )
MAX(expression) MAX(expression)
MEDIAN(median_expression) PERCENTILE_CONT( median_expression, 0.5 ) OVER()
MIN(expression) MIN(expression)
PERCENTILE_CONT(
percentile
) WITHIN GROUP (ORDER BY expression)		 PERCENTILE_CONT(
median_expression,
percentile
) OVER()

Note: Does not cover aggregation use cases.
STDDEV([DISTINCT] expression) STDDEV([DISTINCT] expression)
STDDEV_SAMP([DISTINCT] expression) STDDEV_SAMP([DISTINCT] expression)
STDDEV_POP([DISTINCT] expression) STDDEV_POP([DISTINCT] expression)
SUM([DISTINCT] expression) SUM([DISTINCT] expression)
VARIANCE([DISTINCT] expression) VARIANCE([DISTINCT] expression)
VAR_SAMP([DISTINCT] expression) VAR_SAMP([DISTINCT] expression)
VAR_POP([DISTINCT] expression) VAR_POP([DISTINCT] expression)

BigQuery also offers the following aggregate, aggregate analytic, and approximate aggregate functions, which do not have a direct analogue in Amazon Redshift:

Bitwise aggregate functions

The following table shows mappings between common Amazon Redshift bitwise aggregate functions with their BigQuery equivalents.

Amazon Redshift BigQuery
BIT_AND(expression) BIT_AND(expression)
BIT_OR(expression) BIT_OR(expression)
BOOL_AND>(expression) LOGICAL_AND(expression)
BOOL_OR(expression) LOGICAL_OR(expression)

BigQuery also offers the following bit-wise aggregate function, which does not have a direct analogue in Amazon Redshift:

Window functions

The following table shows mappings between common Amazon Redshift window functions with their BigQuery equivalents. Windowing functions in BigQuery include analytic aggregate functions, aggregate functions, navigation functions, and numbering functions.


Amazon Redshift BigQuery
AVG(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list frame_clause]
)		 AVG(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
COUNT(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list
frame_clause]
)		 COUNT(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
CUME_DIST() OVER
(
[PARTITION BY partition_expression]
[ORDER BY order_list]
) 		CUME_DIST() OVER
(
[PARTITION BY partition_expression]
ORDER BY order_list
)
DENSE_RANK() OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
) 		DENSE_RANK() OVER
(
[PARTITION BY expr_list]
ORDER BY order_list
)
FIRST_VALUE(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list
frame_clause]
)		 FIRST_VALUE(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
LAST_VALUE(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list
frame_clause]
)		 LAST_VALUE(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list
frame_clause]
)
LAG(value_expr [, offset]) OVER
(
[PARTITION BY window_partition]
ORDER BY window_ordering
) 		LAG(value_expr [, offset]) OVER
(
[PARTITION BY window_partition]
ORDER BY window_ordering
)
LEAD(value_expr [, offset]) OVER
(
[PARTITION BY window_partition]
ORDER BY window_ordering
) 		LEAD(value_expr [, offset]) OVER
(
[PARTITION BY window_partition]
ORDER BY window_ordering
)
LISTAGG(
[DISTINCT] expression
[, delimiter]
)
[WITHIN GROUP
(ORDER BY order_list)]
OVER (
[PARTITION BY partition_expression] )		 STRING_AGG(
[DISTINCT] aggregate_expression
[, delimiter] )
OVER (
[PARTITION BY partition_list]
[ORDER BY order_list] )
MAX(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list
frame_clause]
)		 MAX(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
MEDIAN(median_expression) OVER
(
[PARTITION BY partition_expression] ) 		PERCENTILE_CONT(
median_expression,
0.5
)
OVER ( [PARTITION BY partition_expression] )
MIN(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list
frame_clause]
)		 MIN(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
NTH_VALUE(expression, offset) OVER ( [PARTITION BY window_partition] [ORDER BY window_ordering frame_clause] ) NTH_VALUE(expression, offset) OVER
(
[PARTITION BY window_partition]
ORDER BY window_ordering
[frame_clause]
)
NTILE(expr) OVER
(
[PARTITION BY expression_list]
[ORDER BY order_list]
) 		NTILE(expr) OVER
(
[PARTITION BY expression_list]
ORDER BY order_list
)
PERCENT_RANK() OVER
(
[PARTITION BY partition_expression]
[ORDER BY order_list]
) 		PERCENT_RANK() OVER
(
[PARTITION BY partition_expression]
ORDER BY order_list
)
PERCENTILE_CONT(percentile)
WITHIN GROUP (ORDER BY expr) OVER
(
[PARTITION BY expr_list] ) 		PERCENTILE_CONT(expr, percentile) OVER
(
[PARTITION BY expr_list] )
PERCENTILE_DISC(percentile) WITHIN GROUP (ORDER BY expr) OVER
(
[PARTITION BY expr_list]
) 		PERCENTILE_DISC(expr, percentile) OVER
(
[PARTITION BY expr_list] )
RANK() OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
)		 RANK() OVER
(
[PARTITION BY expr_list]
ORDER BY order_list
)
RATIO_TO_REPORT(ratio_expression) OVER
(
[PARTITION BY partition_expression] ) 		ratio_expression SUM(ratio_expression) OVER
(
[PARTITION BY partition_expression] )
ROW_NUMBER() OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
) 		ROW_NUMBER() OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
)
STDDEV(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list
frame_clause]
)		 STDDEV(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause] )
STDDEV_SAMP(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list
frame_clause]
)		 STDDEV_SAMP(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
STDDEV_POP(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
) 		STDDEV_POP(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause] )
SUM(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
) 		SUM(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
VAR_POP(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)		 VAR_POP(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
VAR_SAMP(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
) 		VAR_SAMP(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)
VARIANCE(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)		 VARIANCE(expression) OVER
(
[PARTITION BY expr_list]
[ORDER BY order_list]
[frame_clause]
)

Conditional expressions

The following table shows mappings between common Amazon Redshift conditional expressions with their BigQuery equivalents.

Amazon Redshift BigQuery
CASEexpression
WHEN value THEN result
[WHEN...]
[ELSE else_result]
END		 CASE expression
WHEN value THEN result
[WHEN...]
[ELSE else_result]
END
COALESCE(expression1[, ...]) COALESCE(expression1[, ...])
DECODE(
expression,
search1, result1
[, search2, result2...]
[, default]
) 		CASE expression
WHEN value1 THEN result1
[WHEN value2 THEN result2]
[ELSE default]
END
GREATEST(value [, ...]) GREATEST(value [, ...])
LEAST(value [, ...]) LEAST(value [, ...])
NVL(expression1[, ...]) COALESCE(expression1[, ...])
NVL2(
expression,
not_null_return_value,
null_return_value
)		 IF(
expression IS NULL,
null_return_value,
not_null_return_value
)
NULLIF(expression1, expression2) NULLIF(expression1, expression2)

BigQuery also offers the following conditional expressions, which do not have a direct analogue in Amazon Redshift:

Date and time functions

The following table shows mappings between common Amazon Redshift date and time functions with their BigQuery equivalents. BigQuery data and time functions include date functions, datetime functions, time functions, and timestamp functions.

Keep in mind that functions that seem identical in Amazon Redshift and BigQuery might return different data types.

Amazon Redshift BigQuery
ADD_MONTHS(
date,
integer
)		 CAST( DATE_ADD(
date,
INTERVAL integer MONTH
)
AS TIMESTAMP
)
timestamptz_or_timestamp AT TIME ZONE timezone PARSE_TIMESTAMP(
"%c%z",
FORMAT_TIMESTAMP(
"%c%z",
timestamptz_or_timestamp,
timezone
)
)

Note: Time zones are used when parsing timestamps or formatting timestamps for display. A string-formatted timestamp might include a time zone, but when BigQuery parses the string, it stores the timestamp in the equivalent UTC time. When a time zone is not explicitly specified, the default time zone, UTC, is used. Time zone names or offset from UTC (-HH:MM) are supported, but time zone abbreviations (such as PDT) are not supported.
CONVERT_TIMEZONE(
[source_timezone],
target_timezone,
timestamp
)		 PARSE_TIMESTAMP(
"%c%z",
FORMAT_TIMESTAMP(
"%c%z",
timestamp,
target_timezone
)
)

Note: source_timezone is UTC in BigQuery.
CURRENT_DATE

Note: Returns start date for the current transaction in the current session time zone (UTC by default). 		CURRENT_DATE()

Note: Returns start date for the current statement in the UTC time zone.
DATE_CMP(date1, date2) CASE
WHEN date1 = date2 THEN 0
WHEN date1 > date2 THEN 1
ELSE -1
END
DATE_CMP_TIMESTAMP(date1, date2) CASE
WHEN date1 = CAST(date2 AS DATE)
THEN 0
WHEN date1 > CAST(date2 AS DATE)
THEN 1
ELSE -1
END
DATE_CMP_TIMESTAMPTZ(date, timestamptz) CASE
WHEN date > DATE(timestamptz)
THEN 1
WHEN date < DATE(timestamptz)
THEN -1
ELSE 0
END
DATE_PART_YEAR(date) EXTRACT(YEAR FROM date)
DATEADD(date_part, interval, date) CAST(
DATE_ADD(
date,
INTERVAL interval datepart
)
AS TIMESTAMP
)
DATEDIFF(
date_part,
date_expression1,
date_expression2
) 		DATE_DIFF(
date_expression1,
date_expression2,
date_part
)
DATE_PART(date_part, date) EXTRACT(date_part FROM date)
DATE_TRUNC('date_part', timestamp) TIMESTAMP_TRUNC(timestamp, date_part)
EXTRACT(date_part FROM timestamp) EXTRACT(date_part FROM timestamp)
GETDATE() PARSE_TIMESTAMP(
"%c",
FORMAT_TIMESTAMP(
"%c",
CURRENT_TIMESTAMP()
)
)
INTERVAL_CMP(
interval_literal1,
interval_literal2
) 		For intervals in Redshift, there are 360 days in a year. In BigQuery, you can use the following user-defined function (UDF) to parse a Redshift interval and translate it to seconds.

CREATE TEMP FUNCTION
parse_interval(interval_literal STRING) AS (
(select sum(case
when unit in ('minutes', 'minute', 'm' )
then num * 60
when unit in ('hours', 'hour', 'h') then num
* 60 * 60
when unit in ('days', 'day', 'd' ) then num
* 60 * 60 * 24
when unit in ('weeks', 'week', 'w') then num
* 60 * 60 * 24 * 7
when unit in ('months', 'month' ) then num *
60 * 60 * 24 * 30
when unit in ('years', 'year') then num * 60
* 60 * 24 * 360
else num
end)
from (
select
cast(regexp_extract(value,
r'^[0-9]*\.?[0-9]+') as numeric) num,
substr(value, length(regexp_extract(value,
r'^[0-9]*\.?[0-9]+')) + 1) unit
from
UNNEST(
SPLIT(
replace(
interval_literal, ' ', ''), ',')) value
)));

To compare interval literals, perform:

IF(
parse_interval(interval_literal1) >
parse_interval(interval_literal2),
1,
IF(
parse_interval(interval_literal1) >
parse_interval(interval_literal2),
-1,
0
)
)
LAST_DAY(date) DATE_SUB(
DATE_ADD(
date,
INTERVAL 1 MONTH
),
INTERVAL 1 DAY
)
MONTHS_BETWEEN(
date1,
date2
) 		DATE_DIFF(
date1,
date2,
MONTH
)
NEXT_DAY(date, day) DATE_ADD(
DATE_TRUNC(
date,
WEEK(day)
),
INTERVAL 1 WEEK
)
SYSDATE

Note: Returns start timestamp for the current transaction in the current session time zone (UTC by default). 		CURRENT_TIMESTAMP()

Note: Returns start timestamp for the current statement in the UTC time zone.
TIMEOFDAY() FORMAT_TIMESTAMP(
"%a %b %d %H:%M:%E6S %E4Y %Z",
CURRENT_TIMESTAMP())
TIMESTAMP_CMP(
timestamp1,
timestamp2
)		 CASE
WHEN timestamp1 = timestamp2
THEN 0
WHEN timestamp1 > timestamp2
THEN 1
ELSE -1
END
TIMESTAMP_CMP_DATE(
timestamp,
date
) 		CASE
WHEN
EXTRACT(
DATE FROM timestamp
) = date
THEN 0
WHEN
EXTRACT(
DATE FROM timestamp) > date
THEN 1
ELSE -1
END
TIMESTAMP_CMP_TIMESTAMPTZ(
timestamp,
timestamptz
)

Note: Redshift compares timestamps in the user session-defined time zone. Default user session time zone is UTC. 		CASE
WHEN timestamp = timestamptz
THEN 0
WHEN timestamp > timestamptz
THEN 1
ELSE -1
END

Note: BigQuery compares timestamps in the UTC time zone.
TIMESTAMPTZ_CMP(
timestamptz1,
timestamptz2
)

Note: Redshift compares timestamps in the user session-defined time zone. Default user session time zone is UTC. 		CASE
WHEN timestamptz1 = timestamptz2
THEN 0
WHEN timestamptz1 > timestamptz2
THEN 1
ELSE -1
END

Note: BigQuery compares timestamps in the UTC time zone.
TIMESTAMPTZ_CMP_DATE(
timestamptz,
date
)

Note: Redshift compares timestamps in the user session-defined time zone. Default user session time zone is UTC. 		CASE
WHEN
EXTRACT(
DATE FROM timestamptz) = date
THEN 0
WHEN
EXTRACT(
DATE FROM timestamptz) > date
THEN 1
ELSE -1
END

Note: BigQuery compares timestamps in the UTC time zone.
TIMESTAMPTZ_CMP_TIMESTAMP(
timestamptz,
Timestamp
)

Note: Redshift compares timestamps in the user session-defined time zone. Default user session time zone is UTC. 		CASE
WHEN timestamp = timestamptz
THEN 0
WHEN timestamp > timestamptz
THEN 1
ELSE -1
END

Note: BigQuery compares timestamps in the UTC time zone.
TIMEZONE(
timezone,
Timestamptz_or_timestamp
) 		PARSE_TIMESTAMP(
"%c%z", FORMAT_TIMESTAMP(
"%c%z",
timestamptz_or_timestamp,
timezone
)
)

Note: Time zones are used when parsing timestamps or formatting timestamps for display. A string-formatted timestamp might include a time zone, but when BigQuery parses the string, it stores the timestamp in the equivalent UTC time. When a time zone is not explicitly specified, the default time zone, UTC, is used. Time zone names or offset from UTC (-HH:MM) are supported but time zone abbreviations (such as PDT) are not supported.
TO_TIMESTAMP(timestamp, format) PARSE_TIMESTAMP(
format,
FORMAT_TIMESTAMP(
format,
timestamp
)
)

Note: BigQuery follows a different set of format elements. Time zones are used when parsing timestamps or formatting timestamps for display. A string-formatted timestamp might include a time zone, but when BigQuery parses the string, it stores the timestamp in the equivalent UTC time. When a time zone is not explicitly specified, the default time zone, UTC, is used. Time zone names or offset from UTC (-HH:MM) are supported in the format string but time zone abbreviations (such as PDT) are not supported.
TRUNC(timestamp) CAST(timestamp AS DATE)

BigQuery also offers the following date and time functions, which do not have a direct analogue in Amazon Redshift:

Mathematical operators

The following table shows mappings between common Amazon Redshift mathematical operators with their BigQuery equivalents.

Amazon Redshift BigQuery

X + Y

X + Y

X - Y

X - Y

X * Y

X * Y

X / Y


Note: If the operator is
performing integer division (in other words, if X and Y are both integers), an integer is returned. If the operator is performing non-integer division, a non-integer is returned.		 If integer division:
CAST(FLOOR(X / Y) AS INT64)

If not integer division:

CAST(X / Y AS INT64)


Note: Division in BigQuery returns a non-integer.
To prevent errors from a division operation (division by zero error), use SAFE_DIVIDE(X, Y) or IEEE_DIVIDE(X, Y).

X % Y

MOD(X, Y)


Note: To prevent errors from a division operation (division by zero error), use SAFE.MOD(X, Y). SAFE.MOD(X, 0) results in 0.

X ^ Y

POW(X, Y)

POWER(X, Y)


Note: Unlike Amazon Redshift, the ^ operator in BigQuery performs Bitwise xor.

| / X

SQRT(X)


Note: To prevent errors from a square root operation (negative input), use SAFE.SQRT(X). Negative input with SAFE.SQRT(X) results in NULL.

|| / X

SIGN(X) * POWER(ABS(X), 1/3)


Note: BigQuery's POWER(X, Y) returns an error if X is a finite value less than 0 and Y is a noninteger.

@ X

ABS(X)

X << Y

X << Y


Note: This operator returns 0 or a byte sequence of b'\x00' if the second operand Y is greater than or equal to the bit length of the first operand X (for example, 64 if X has the type INT64). This operator throws an error if Y is negative.

X >> Y

X >> Y


Note: Shifts the first operand X to the right. This operator does not do sign bit extension with a signed type (it fills vacant bits on the left with 0). This operator returns 0 or a byte sequence of
b'\x00' if the second operand Y is greater than or equal to the bit length of the first operand X (for example, 64 if X has the type INT64). This operator throws an error if Y is negative.

X & Y

X & Y

X | Y

X | Y

~X

~X

BigQuery also offers the following mathematical operator, which does not have a direct analog in Amazon Redshift:

Math functions

Amazon Redshift BigQuery
ABS(number) ABS(number)
ACOS(number) ACOS(number)
ASIN(number) ASIN(number)
ATAN(number) ATAN(number)
ATAN2(number1, number2) ATAN2(number1, number2)
CBRT(number) POWER(number, 1/3)
CEIL(number) CEIL(number)
CEILING(number) CEILING(number)
CHECKSUM(expression) FARM_FINGERPRINT(expression)
COS(number) COS(number)
COT(number) 1/TAN(number)
DEGREES(number) number*180/ACOS(-1)
DEXP(number) EXP(number)
DLOG1(number) LN(number)
DLOG10(number) LOG10(number)
EXP(number) EXP(number)
FLOOR(number) FLOOR(number)
LNnumber) LN(number)
LOG(number) LOG10(number)
MOD(number1, number2) MOD(number1, number2)
PI ACOS(-1)
POWER(expression1, expression2) POWER(expression1, expression2)
RADIANS(number) ACOS(-1)*(number/180)
RANDOM() RAND()
ROUND(number [, integer]) ROUND(number [, integer])
SIN(number) SIN(number)
SIGN(number) SIGN(number)
SQRT(number) SQRT(number)
TAN(number) TAN(number)
TO_HEX(number) FORMAT('%x', number)
TRUNC(number [, integer])+-+++ TRUNC(number [, integer])

String functions

Amazon Redshift BigQuery
string1 || string2 CONCAT(string1, string2)
BPCHARCMP(string1, string2) CASE
WHEN string1 = string2 THEN 0
WHEN string1 > string2 THEN 1
ELSE -1
END
BTRIM(string [, matching_string]) TRIM(string [, matching_string])
BTTEXT_PATTERN_CMP(string1, string2) CASE
WHEN string1 = string2 THEN 0
WHEN string1 > string2 THEN 1
ELSE -1
END
CHAR_LENGTH(expression) CHAR_LENGTH(expression)
CHARACTER_LENGTH(expression) CHARACTER_LENGTH(expression)
CHARINDEX(substring, string) STRPOS(string, substring)
CHR(number) CODE_POINTS_TO_STRING([number])
CONCAT(string1, string2) CONCAT(string1, string2)

Note: BigQuery's CONCAT(...) supports
concatenating any number of strings.
CRC32 Custom user-defined function
FUNC_SHA1(string) SHA1(string)
INITCAP INITCAP
LEFT(string, integer) SUBSTR(string, 0, integer)
RIGHT(string, integer) SUBSTR(string, -integer)
LEN(expression) LENGTH(expression)
LENGTH(expression) LENGTH(expression)
LOWER(string) LOWER(string)
LPAD(string1, length[, string2]) LPAD(string1, length[, string2])
RPAD(string1, length[, string2]) RPAD(string1, length[, string2])
LTRIM(string, trim_chars) LTRIM(string, trim_chars)
MD5(string) MD5(string)
OCTET_LENGTH(expression) BYTE_LENGTH(expression)
POSITION(substring IN string) STRPOS(string, substring)
QUOTE_IDENT(string) CONCAT('"',string,'"')
QUOTE_LITERAL(string) CONCAT("'",string,"'")
REGEXP_COUNT( source_string, pattern
[,position]
) 		ARRAY_LENGTH( REGEXP_EXTRACT_ALL(
source_string,
pattern
)
)

If position is specified:

ARRAY_LENGTH( REGEXP_EXTRACT_ALL(
SUBSTR(source_string, IF(position <= 0, 1, position)),
pattern
)
)

Note: BigQuery provides regular expression support using the re2 library; see that documentation for its regular expression syntax.
REGEXP_INSTR(
source_string,
pattern
[,position
[,occurrence]] ) 		IFNULL( STRPOS(
source_string, REGEXP_EXTRACT(
source_string,
pattern)
),0)

If source_string is specified:

REGEXP_REPLACE(
source_string,

pattern,
replace_string
)

If position is specified:

IFNULL( STRPOS(
SUBSTR(source_string, IF(position
<= 0, 1, position)), REGEXP_EXTRACT(
SUBSTR(source_string, IF(position <= 0, 1, position)),
pattern)
) + IF(position <= 0, 1, position) - 1, 0)

If occurrence is specified:

IFNULL( STRPOS(
SUBSTR(source_string, IF(position
<= 0, 1, position)), REGEXP_EXTRACT_ALL(
SUBSTR(source_string, IF(position <= 0, 1, position)),
pattern
)[SAFE_ORDINAL(occurrence)]
) + IF(position <= 0, 1, position) - 1, 0)

Note: BigQuery provides regular expression
support using the re2 library; see that
documentation for its regular expression
syntax.
REGEXP_REPLACE( source_string,
pattern
[, replace_string [, position]]
) 		REGEXP_REPLACE(
source_string,
pattern,
""
)

If source_string is specified:

REGEXP_REPLACE(
source_string,

pattern, replace_string
)

If position is specified:

CASE
WHEN position > LENGTH(source_string) THEN source_string
WHEN position <= 0 THEN REGEXP_REPLACE(
source_string, pattern,
""
) ELSE
CONCAT( SUBSTR(
source_string, 1, position - 1), REGEXP_REPLACE(
SUBSTR(source_string, position), pattern,
replace_string
)
) END
REGEXP_SUBSTR( source_string, pattern
[, position
[, occurrence]] ) 		REGEXP_EXTRACT(
source_string, pattern
)

If position is specified:

REGEXP_EXTRACT(
SUBSTR(source_string, IF(position <= 0, 1, position)),
pattern

)

If occurrence is specified:

REGEXP_EXTRACT_ALL(
SUBSTR(source_string, IF(position <= 0, 1, position)),


pattern
)[SAFE_ORDINAL(occurrence)]

Note: BigQuery provides regular expression support using the re2 library; see that documentation for its regular expression syntax.
REPEAT(string, integer) REPEAT(string, integer)
REPLACE(string, old_chars, new_chars) REPLACE(string, old_chars, new_chars)
REPLICA(string, integer) REPEAT(string, integer)
REVERSE(expression) REVERSE(expression)
RTRIM(string, trim_chars) RTRIM(string, trim_chars)
SPLIT_PART(string, delimiter, part) SPLIT(
string
delimiter
)SAFE_ORDINAL(part)
STRPOS(string, substring) STRPOS(string, substring)
STRTOL(string, base)
SUBSTRING(
string,
start_position, number_characters ) 		SUBSTR(
string,
start_position, number_characters )
TEXTLEN(expression) LENGTH(expression)
TRANSLATE(
expression,
characters_to_replace, characters_to_substitute ) 		Can be implemented using UDFs:

CREATE TEMP FUNCTION
translate(expression STRING,
characters_to_replace STRING, characters_to_substitute STRING) AS ( IF(LENGTH(characters_to_replace) < LENGTH(characters_to_substitute) OR LENGTH(expression) <
LENGTH(characters_to_replace), expression,
(SELECT
STRING_AGG(
IFNULL(
(SELECT ARRAY_CONCAT([c],
SPLIT(characters_to_substitute, ''))[SAFE_OFFSET((
SELECT IFNULL(MIN(o2) + 1,
0) FROM
UNNEST(SPLIT(characters_to_replace,
'')) AS k WITH OFFSET o2
WHERE k = c))]
),
''),
'' ORDER BY o1)
FROM UNNEST(SPLIT(expression, ''))
AS c WITH OFFSET o1
))
);
TRIM([BOTH] string) TRIM(string)
TRIM([BOTH] characters FROM string) TRIM(string, characters)
UPPER(string) UPPER(string)

Data type formatting functions

Amazon Redshift BigQuery
CAST(expression AS type) CAST(expression AS type)
expression :: type CAST(expression AS type)
CONVERT(type, expression) CAST(expression AS type)
TO_CHAR(
timestamp_expression, format
)		 FORMAT_TIMESTAMP(
format,
timestamp_expression
)

Note: BigQuery and Amazon Redshift differ in how to specify a format string for timestamp_expression.
TO_CHAR(
numeric_expression,
format
)		 FORMAT(
format,
numeric_expression
)

Note: BigQuery and Amazon Redshift differ in how to specify a format string for timestamp_expression.
TO_DATE(date_string, format) PARSE_DATE(date_string, format)

Note: BigQuery and Amazon Redshift differ in how to specify a format string for date_string.
TO_NUMBER(string, format) CAST(
FORMAT(
format,
numeric_expression
) TO INT64
)

Note: BigQuery and Amazon Redshift differ in how to specify a numeric format string.

BigQuery also supports SAFE_CAST(expression AS typename), which returns NULL if BigQuery is unable to perform a cast; for example, SAFE_CAST("apple" AS INT64) returns NULL.

DML syntax

This section addresses differences in data management language syntax between Amazon Redshift and BigQuery.

INSERT statement

Amazon Redshift offers a configurable DEFAULT keyword for columns. In BigQuery, the DEFAULT value for nullable columns is NULL, and DEFAULT is not supported for required columns. Most Amazon Redshift INSERT statements are compatible with BigQuery. The following table shows exceptions.

Amazon Redshift BigQuery
INSERT INTO table (column1 [, ...])
DEFAULT VALUES		 INSERT [INTO] table (column1 [, ...])
VALUES (DEFAULT [, ...])
INSERT INTO table (column1, [,...]) VALUES (
SELECT ...
FROM ...
)		 INSERT [INTO] table (column1, [,...])
SELECT ...
FROM ...

BigQuery also supports inserting values using a subquery (where one of the values is computed using a subquery), which is not supported in Amazon Redshift. For example:

INSERT INTO table (column1, column2)
VALUES ('value_1', (
SELECT column2
FROM table2
))

COPY statement

Amazon Redshift's COPYcommand loads data into a table from data files or from an Amazon DynamoDB table. BigQuery does not use the SQL COPY command to load data, but you can use any of several non-SQL tools and options to load data into BigQuery tables. You can also use data pipeline sinks provided in Apache Spark or Apache Beam to write data into BigQuery.

UPDATE statement

Most Amazon Redshift UPDATE statements are compatible with BigQuery. The following table shows exceptions.

Amazon Redshift BigQuery
UPDATE table
SET column = expression [,...] [FROM ...] 		UPDATE table
SET column = expression [,...]
[FROM ...]
WHERE TRUE

Note: All UPDATE statements in BigQuery require a WHERE keyword, followed by a condition.
UPDATE table
SET column = DEFAULT [,...] [FROM ...]
[WHERE ...]		 UPDATE table
SET column = NULL [, ...]
[FROM ...]
WHERE ...

Note: BigQuery's UPDATE command does not support DEFAULT values.

If the Amazon Redshift UPDATE statement does not include a WHERE clause, the BigQuery UPDATE statement should be conditioned WHERE TRUE.

DELETE and TRUNCATE statements

The DELETE and TRUNCATE statements are both ways to remove rows from a table without affecting the table schema or indexes.

In Amazon Redshift, the TRUNCATE statement is recommended over an unqualified DELETE statement because it is faster and does not require VACUUM and ANALYZE operations afterward. However, you can use DELETE statements to achieve the same effect.

In BigQuery, the DELETE statement must have a WHERE clause. For more information about DELETE in BigQuery, see the BigQuery DELETE examples in the DML documentation.

Amazon Redshift BigQuery
DELETE [FROM] table_name

TRUNCATE [TABLE] table_name		 DELETE FROM table_name
WHERE TRUE

BigQuery DELETE statements require a WHERE clause.
DELETE FROM table_name
USING other_table
WHERE table_name.id=other_table.id		 DELETE FROM table_name
WHERE table_name.id IN (
SELECT id
FROM other_table
)

DELETE FROM table_name
WHERE EXISTS (
SELECT id
FROM other_table
WHERE table_name.id = other_table.id )

In Amazon Redshift, USING allows additional tables to be referenced in the WHERE clause. This can be achieved in BigQuery by using a subquery in the WHERE clause.

MERGE statement

The MERGE statement can combine INSERT, UPDATE, and DELETE operations into a single upsert statement and perform the operations atomically. The MERGE operation must match at most one source row for each target row.

Amazon Redshift does not support a single MERGE command. However, a merge operation can be performed in Amazon Redshift by performing INSERT, UPDATE, and DELETE operations in a transaction.

Merge operation by replacing existing rows

In Amazon Redshift, an overwrite of all of the columns in the target table can be performed using a DELETE statement and then an INSERT statement. The DELETE statement removes rows that should be updated, and then the INSERT statement inserts the updated rows. BigQuery tables are limited to 1,000 DML statements per day, so you should consolidate INSERT, UPDATE, and DELETE statements into a single MERGE statement as shown in the following table.

Amazon Redshift BigQuery
See Performing a merge operation by replacing existing rows.

CREATE TEMP TABLE temp_table;

INSERT INTO temp_table
SELECT *
FROM source
WHERE source.filter = 'filter_exp';

BEGIN TRANSACTION;

DELETE FROM target
USING temp_table
WHERE target.key = temp_table.key;

INSERT INTO target
SELECT *
FROM temp_table;

END TRANSACTION;

DROP TABLE temp_table; 		MERGE target
USING source
ON target.key = source.key
WHEN MATCHED AND source.filter = 'filter_exp' THEN
UPDATE SET
target.col1 = source.col1,
target.col2 = source.col2,
...

Note: All columns must be listed if updating all columns.
See Performing a merge operation by specifying a column list.

CREATE TEMP TABLE temp_table;

INSERT INTO temp_table
SELECT *
FROM source
WHERE source.filter = 'filter_exp';

BEGIN TRANSACTION;

UPDATE target SET
col1 = temp_table.col1,
col2 = temp_table.col2
FROM temp_table
WHERE target.key=temp_table.key;

INSERT INTO target
SELECT *
FROM 		MERGE target
USING source
ON target.key = source.key
WHEN MATCHED AND source.filter = 'filter_exp' THEN
UPDATE SET
target.col1 = source.col1,
target.col2 = source.col2

DDL syntax

This section addresses differences in data definition language syntax between Amazon Redshift and BigQuery.

SELECT INTO statement

In Amazon Redshift, the SELECT INTO statement can be used to insert the results of a query into a new table, combining table creation and insertion.

Amazon Redshift BigQuery
SELECT expression, ... INTO table
FROM ... 		INSERT table
SELECT expression, ...
FROM ...
WITH subquery_table AS ( SELECT ...
)
SELECT expression, ... INTO table
FROM subquery_table
... 		INSERT table
WITH subquery_table AS (
SELECT ...
)
SELECT expression, ...
FROM subquery_table
...
SELECT expression
INTO TEMP table
FROM ...

SELECT expression
INTO TEMPORARY table
FROM ... 		BigQuery offers several ways to emulate temporary tables. See the temporary tables section for more information.

CREATE TABLE statement

Most Amazon Redshift CREATE TABLE statements are compatible with BigQuery, except for the following syntax elements, which are not used in BigQuery:

Amazon Redshift BigQuery
CREATE TABLE table_name (
col1 data_type1 NOT NULL,
col2 data_type2 NULL,
col3 data_type3 UNIQUE,
col4 data_type4 PRIMARY KEY,
col5 data_type5
)

Note: UNIQUE and PRIMARY KEY constraints are informational and are not enforced by the Amazon Redshift system.		 CREATE TABLE table_name (
col1 data_type1 NOT NULL,
col2 data_type2,
col3 data_type3,
col4 data_type4,
col5 data_type5,
)
CREATE TABLE table_name
(
col1 data_type1[,...]
table_constraints
)
where table_constraints are:
[UNIQUE(column_name [, ... ])]
[PRIMARY KEY(column_name [, ...])]
[FOREIGN KEY(column_name [, ...])
REFERENCES reftable [(refcolumn)]

Note: UNIQUE and PRIMARY KEY constraints are informational and are not enforced by the Amazon Redshift system. 		CREATE TABLE table_name
(
col1 data_type1[,...]
)
PARTITION BY column_name
CLUSTER BY column_name [, ...]

Note: BigQuery does not use UNIQUE, PRIMARY KEY, or FOREIGN KEY table constraints. To achieve similar optimization that these constraints provide during query execution, partition and cluster your BigQuery tables. CLUSTER BY supports up to 4 columns.
CREATE TABLE table_name
LIKE original_table_name 		Reference this example to learn how to use the INFORMATION_SCHEMA tables to copy column names, data types, and NOT NULL constraints to a new table.
CREATE TABLE table_name
(
col1 data_type1
)
BACKUP NO

Note: In Amazon Redshift, the BACKUP NO setting is specified to save processing time and reduce storage space.		 The BACKUP NO table option is not used or needed because BigQuery automatically keeps up to 7 days of historical versions of all of your tables with no effect on processing time or billed storage.
CREATE TABLE table_name
(
col1 data_type1
)
table_attributes
where table_attributes are:
[DISTSTYLE {AUTO|EVEN|KEY|ALL}]
[DISTKEY (column_name)]
[[COMPOUND|INTERLEAVED] SORTKEY
(column_name [, ...])] 		BigQuery supports clustering, which allows storing keys in sorted order.
CREATE TABLE table_name
AS SELECT ... 		CREATE TABLE table_name
AS SELECT ...
CREATE TABLE IF NOT EXISTS table_name ... CREATE TABLE IF NOT EXISTS
table_name
...

BigQuery also supports the DDL statement CREATE OR REPLACE TABLE, which overwrites a table if it already exists.

BigQuery's CREATE TABLE statement also supports the following clauses, which do not have an Amazon Redshift equivalent:

For more information about CREATE TABLE in BigQuery, see the BigQuery CREATE TABLE examples in the DML documentation.

Temporary tables

Amazon Redshift supports temporary tables, which are only visible within the current session. There are several ways to emulate temporary tables in BigQuery:

  • Dataset TTL: Create a dataset that has a short time to live (for example, one hour) so that any tables created in the dataset are effectively temporary because they won't persist longer than the dataset's time to live. You can prefix all of the table names in this dataset with temp to clearly denote that the tables are temporary.

  • Table TTL: Create a table that has a table-specific short time to live using DDL statements similar to the following:

    CREATE TABLE
temp.name (col1, col2, ...)
OPTIONS (expiration_timestamp=TIMESTAMP_ADD(CURRENT_TIMESTAMP(),
INTERVAL 1 HOUR));

CREATE VIEW statement

The following table shows equivalents between Amazon Redshift and BigQuery for the CREATE VIEW statement.

Amazon Redshift BigQuery
CREATE VIEW view_name AS SELECT ...code> CREATE VIEW view_name AS SELECT ...
CREATE OR REPLACE VIEW view_name AS SELECT ... CREATE OR REPLACE VIEW
view_name AS
SELECT ...
CREATE VIEW view_name
(column_name, ...)
AS SELECT ... 		CREATE VIEW view_name AS SELECT ...
Not supported. CREATE VIEW IF NOT EXISTS c view_name
OPTIONS(view_option_list)
AS SELECT …

Creates a new view only if the view does not exist in the specified dataset.
CREATE VIEW view_name
AS SELECT ...
WITH NO SCHEMA BINDING

In Amazon Redshift, a late binding view is required in order to reference an external table.		 In BigQuery, to create a view, all referenced objects must already exist.

BigQuery allows you to query external data sources.

User-defined functions (UDFs)

A UDF lets you create functions for custom operations. These functions accept columns of input, perform actions, and return the result of those actions as a value.

Both Amazon Redshift and BigQuery support UDFs using SQL expressions. Additionally, in Amazon Redshift you can create a Python-based UDF, and in BigQuery you can create a JavaScript-based UDF.

Refer to the Google Cloud BigQuery utilities GitHub repository for a library of common BigQuery UDFs.

CREATE FUNCTION syntax

The following table addresses differences in SQL UDF creation syntax between Amazon Redshift and BigQuery.

Amazon Redshift BigQuery
CREATE [OR REPLACE] FUNCTION
function_name
([sql_arg_name sql_arg_data_type[,..]]) RETURNS data_type
IMMUTABLE
AS $$
sql_function_definition
$$ LANGUAGE sql 		CREATE [OR REPLACE] FUNCTION
function_name
([sql_arg_name sql_arg_data_type[,..]]) AS
sql_function_definition

Note: In a BigQuery SQL UDF, a return data type is optional. BigQuery infers the result type of the function from the SQL function body when a query calls the function.
CREATE [OR REPLACE] FUNCTION
function_name
([sql_arg_name sql_arg_data_type[,..]]) RETURNS data_type
{ VOLATILE | STABLE | IMMUTABLE } AS $$
sql_function_definition
$$ LANGUAGE sql 		CREATE [OR REPLACE] FUNCTION
function_name
([sql_arg_name sql_arg_data_type[,..]]) RETURNS data_type
AS sql_function_definition

Note: Function volatility is not a configurable parameter in BigQuery. All BigQuery UDF volatility is equivalent to Amazon Redshift's IMMUTABLE volatility (that is, it does not do database lookups or otherwise use information not directly present in its argument list).
CREATE [OR REPLACE] FUNCTION
function_name
([sql_arg_name sql_arg_data_type[,..]]) RETURNS data_type
IMMUTABLE
AS $$
SELECT_clause
$$ LANGUAGE sql

Note: Amazon Redshift supports only a SQL SELECT clause as function definition. Also, the SELECT clause cannot include any of the FROM, INTO, WHERE, GROUP BY, ORDER BY, and LIMIT clauses. 		CREATE [OR REPLACE] FUNCTION
function_name
([sql_arg_name sql_arg_data_type[,..]]) RETURNS data_type
AS sql_expression

Note: BigQuery supports any SQL expressions as function definition. However, referencing tables, views, or models is not supported.
CREATE [OR REPLACE] FUNCTION
function_name
([sql_arg_name sql_arg_data_type[,..]]) RETURNS data_type
IMMUTABLE
AS $$
sql_function_definition
$$ LANGUAGE sql 		CREATE [OR REPLACE] FUNCTION function_name ([sql_arg_name sql_arg_data_type[,..]]) RETURNS data_type AS sql_function_definition

Note: Language literal need not be specified in a GoogleSQL UDF. BigQuery interprets the SQL expression by default. Also, the Amazon Redshift dollar quoting ($$) is not supported in BigQuery.
CREATE [OR REPLACE] FUNCTION function_name (integer, integer) RETURNS integer IMMUTABLE AS $$ SELECT $1 + $2 $$ LANGUAGE sql CREATE [OR REPLACE] FUNCTION
function_name
(x INT64, y INT64)
RETURNS INT64
AS
SELECT x + y

Note: BigQuery UDFs require all input arguments to be named. The Amazon Redshift argument variables ($1, $2, …) are not supported in BigQuery.
CREATE [OR REPLACE] FUNCTION
function_name
(integer, integer)
RETURNS integer
IMMUTABLE
AS $$
SELECT $1 + $2
$$ LANGUAGE sql

Note: Amazon Redshift does not support ANY TYPE for SQL UDFs. However, it supports using the ANYELEMENT data type in Python-based UDFs. 		CREATE [OR REPLACE] FUNCTION
function_name
(x ANY TYPE, y ANY TYPE)
AS
SELECT x + y

Note: BigQuery supports using ANY TYPE as argument type. The function accepts an input of any type for this argument. For more information, see templated parameter in BigQuery.

BigQuery also supports the CREATE FUNCTION IF NOT EXISTS statement, which treats the query as successful and takes no action if a function with the same name already exists.

BigQuery's CREATE FUNCTION statement also supports creating TEMPORARY or TEMP functions, which do not have an Amazon Redshift equivalent.

See calling UDFs for details on executing a BigQuery-persistent UDF.

DROP FUNCTION syntax

The following table addresses differences in DROP FUNCTION syntax between Amazon Redshift and BigQuery.

Amazon Redshift BigQuery
DROP FUNCTION
function_name
( [arg_name] arg_type [, ...] ) [ CASCADE | RESTRICT ]		 DROP FUNCTION
dataset_name.function_name

Note: BigQuery does not require using the function's signature for deleting the function. Also, removing function dependencies is not supported in BigQuery.

BigQuery also supports the DROP FUNCTION IF EXISTS statement, which deletes the function only if the function exists in the specified dataset.

BigQuery requires that you specify the project_name if the function is not located in the current project.

UDF components

This section highlights the similarities and differences in UDF components between Amazon Redshift andBigQuery.

Component Amazon Redshift BigQuery
Name Amazon Redshift recommends using the prefix _f for function names to avoid conflicts with existing or future built-in SQL function names. In BigQuery, you can use any custom function name.
Arguments Arguments are optional. You can use name and data types for Python UDF arguments and only data types for SQL UDF arguments. In SQL UDFs, you must refer to arguments using $1, $2, and so on. Amazon Redshift also restricts the number of arguments to 32. Arguments are optional, but if you specify arguments, they must use both name and data types for both JavaScript and SQL UDFs. The maximum number of arguments for a persistent UDF is 256.
Data type Amazon Redshift supports a different set of data types for SQL and Python UDFs.
For a Python UDF, the data type might also be ANYELEMENT.

You must specify a RETURN data type for both SQL and Python UDFs.

See Data types in this document for equivalents between data types in Amazon Redshift and in BigQuery.		 BigQuery supports a different set of data types for SQL and JavaScript UDFs.
For a SQL UDF, the data type might also be ANY TYPE. For more information, see templated parameters in BigQuery.

The RETURN data type is optional for SQL UDFs.

See SQL type encodings in JavaScript for information on how BigQuery data types map to JavaScript data types.
Definition For both SQL and Python UDFs, you must enclose the function definition using dollar quoting, as in a pair of dollar signs ($$), to indicate the start and end of the function statements.

For SQL UDFs, Amazon Redshift supports only a SQL SELECT clause as the function definition. Also, the SELECT clause cannot include any of the FROM, INTO, WHERE, GROUP
BY, ORDER BY, and LIMIT clauses.

For Python UDFs, you can write a Python program using the Python 2.7 Standard Library or import your custom modules by creating one using the CREATE LIBRARY command.		 In BigQuery, you need to enclose the JavaScript code in quotes. See Quoting rules for more information.

For SQL UDFs, you can use any SQL expressions as the function definition. However, BigQuery doesn't support referencing tables, views, or models.

For JavaScript UDFs, you can include external code libraries directly using the OPTIONS section. You can also use the BigQuery UDF test tool to test your functions.
Language You must use the LANGUAGE literal to specify the language as either sql for SQL UDFs or plpythonu for Python UDFs. You need not specify LANGUAGE for SQL UDFs but must specify the language as js for JavaScript UDFs.
State Amazon Redshift does not support creating temporary UDFs.

Amazon Redshift provides an option to define the volatility of a function using VOLATILE, STABLE, or IMMUTABLE literals. This is used for optimization by the query optimizer.		 BigQuery supports both persistent and temporary UDFs. You can reuse persistent UDFs across multiple queries, whereas you can only use temporary UDFs in a single query.

Function volatility is not a configurable parameter in BigQuery. All BigQuery UDF volatility is equivalent to Amazon Redshift's IMMUTABLE volatility.
Security and privileges To create a UDF, you must have permission for usage on language for SQL or plpythonu (Python). By default, USAGE ON LANGUAGE SQL is granted to PUBLIC, but you must explicitly grant USAGE ON LANGUAGE PLPYTHONU to specific users or groups.
Also, you must be a superuser to replace a UDF.		 Granting explicit permissions for creating or deleting any type of UDF is not necessary in BigQuery. Any user assigned a role of BigQuery Data Editor (having bigquery.routines.* as one of the permissions) can create or delete functions for the specified dataset.

BigQuery also supports creating custom roles. This can be managed using Cloud IAM.
Limits See Python UDF limits. See UDF limits.

Metadata and transaction SQL statements

Amazon Redshift BigQuery
SELECT * FROM STL_ANALYZE WHERE name
= 'T';		 Not used in BigQuery. You don't need to gather statistics in order to improve query performance. To get information about your data distribution, you can use approximate aggregate functions.
ANALYZE [[ table_name[(column_name
[, ...])]]		 Not used in BigQuery.
LOCK TABLE table_name; Not used in BigQuery.
BEGIN TRANSACTION; SELECT ...
END TRANSACTION;		 BigQuery uses snapshot isolation. For details, see Consistency guarantees.
EXPLAIN ... Not used in BigQuery.

Similar features are the query plan explanation in the BigQuery Google Cloud console, and in audit logging in Cloud Monitoring.
SELECT * FROM SVV_TABLE_INFO WHERE
table = 'T';		 SELECT * EXCEPT(is_typed) FROM
mydataset.INFORMATION_SCHEMA.TABLES;

For more information see Introduction to BigQuery INFORMATION_SCHEMA.
VACUUM [table_name] Not used in BigQuery. BigQuery clustered tables are automatically sorted.

Multi-statement and multi-line SQL statements

Both Amazon Redshift and BigQuery support transactions (sessions) and therefore support statements separated by semicolons that are consistently executed together. For more information, see Multi-statement transactions.

Procedural SQL statements

CREATE PROCEDURE statement

Amazon Redshift BigQuery
CREATE or REPLACE PROCEDURE CREATE PROCEDURE if a name is required.

Otherwise, use inline with BEGIN or in a single line with CREATE TEMP FUNCTION.
CALL CALL

Variable declaration and assignment

Amazon Redshift BigQuery
DECLARE DECLARE

Declares a variable of the specified type.
SET SET

Sets a variable to have the value of the provided expression, or sets multiple variables at the same time based on the result of multiple expressions.

Error condition handlers

In Amazon Redshift, an error encountered during the execution of a stored procedure ends the execution flow, ends the transaction, and rolls back the transaction. These results occur because subtransactions are not supported. In an Amazon Redshift-stored procedure, the only supported handler_statement is RAISE. In BigQuery, error handling is a core feature of the main control flow, similar to what other languages provide with TRY ... CATCH blocks.

Amazon Redshift BigQuery
BEGIN ... EXCEPTION WHEN OTHERS THEN BEGIN ... EXCEPTION WHEN ERROR THEN
RAISE RAISE
[ <<label>> ] [ DECLARE declarations ]
BEGIN
statements EXCEPTION
BEGIN
statements
EXCEPTION
WHEN OTHERS THEN
Handler_statements
END;		 BEGIN
BEGIN
...
EXCEPTION WHEN ERROR THEN SELECT 1/0;
END;

EXCEPTION WHEN ERROR THEN -- The exception thrown from the inner exception handler lands here. END;

Cursor declarations and operations

Because BigQuery doesn't support cursors or sessions, the following statements aren't used in BigQuery:

If you're using the cursor to return a result set, you can achieve similar behavior using temporary tables in BigQuery.

Dynamic SQL statements

The scripting feature in BigQuery supports dynamic SQL statements like those shown in the following table.

Amazon Redshift BigQuery
EXECUTE EXECUTE IMMEDIATE

Flow-of-control statements

Amazon Redshift BigQuery
IF..THEN..ELSIF..THEN..ELSE..END IF IF condition
THEN stmts
ELSE stmts
END IF
name CURSOR [ ( arguments ) ] FOR query Cursors or sessions are not used in BigQuery.
[< LOOP
sql_statement_list END LOOP;
WHILE condition LOOP stmts END LOOP WHILE condition
DO stmts
END WHILE
EXIT BREAK

Consistency guarantees and transaction isolation

Both Amazon Redshift and BigQuery are atomic—that is, ACID-compliant on a per-mutation level across many rows.

Transactions

Amazon Redshift supports serializable isolation by default for transactions. Amazon Redshift lets you specify any of the four SQL standard transaction isolation levels but processes all isolation levels as serializable.

BigQuery also supports transactions. BigQuery helps ensure optimistic concurrency control (first to commit has priority) with snapshot isolation, in which a query reads the last committed data before the query starts. This approach guarantees the same level of consistency on a per-row, per-mutation basis and across rows within the same DML statement, yet avoids deadlocks. In the case of multiple DML updates against the same table, BigQuery switches to pessimistic concurrency control. Load jobs can run completely independently and append to tables.

Rollback

If Amazon Redshift encounters any error while running a stored procedure, it rolls back all changes made in a transaction. Additionally, you can use the ROLLBACK transaction control statement in a stored procedure to discard all changes.

In BigQuery, you can use the ROLLBACK TRANSACTION statement.

Database limits

Check the BigQuery public documentation for the latest quotas and limits. Many quotas for large-volume users can be raised by contacting the Cloud support team. The following table shows a comparison of the Amazon Redshift and BigQuery database limits.

Limit Amazon Redshift BigQuery
Tables in each database for large and xlarge cluster node types 9,900 Unrestricted
Tables in each database for 8xlarge cluster node types 20,000 Unrestricted
User-defined databases you can create for each cluster 60 Unrestricted
Maximum row size 4 MB 100 MB