DELIMITER $$

DROP FUNCTION IF EXISTS `fac201001`.`nextval` $$
CREATE FUNCTION `sequence`.`nextval` (`seq_name` varchar(100)) RETURNS bigint(20) NOT DETERMINISTIC
BEGIN
DECLARE cur_val bigint(20);

SELECT
sequence_cur_value INTO cur_val
FROM
`sequence`.`sequence_data`
WHERE
sequence_name = seq_name;

IF cur_val IS NOT NULL THEN
UPDATE
`sequence`.`sequence_data`
SET
sequence_cur_value = IF (
(sequence_cur_value + sequence_increment) > sequence_max_value,
IF (
sequence_cycle = TRUE,
sequence_min_value,
NULL
),
sequence_cur_value + sequence_increment
)
WHERE
sequence_name = seq_name;
END IF;

RETURN cur_val;

END $$

DELIMITER ;

Another thing is that using one table for each sequence, plus if your sequence is large (for example the current sequence is 1M), then you essentially have a table with 1M rows. That would be overkill, I think.

A table to represent each sequence: (In this example, a batch ID sequence)

DROP TABLE IF EXISTS BatchIDSequence;

CREATE TABLE BatchIDSequence (
ID int(11) NOT NULL AUTO_INCREMENT
,processID int(11) NOT NULL
,PRIMARY KEY (ID)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;

A NextVal stored procedure to that uses a prepared statement and accepts the sequence name as the parameter.

DROP PROCEDURE IF EXISTS pNextVal;
CREATE PROCEDURE pNextVal(in_sequenceName VARCHAR(60))
BEGIN
SET @stmt_text=CONCAT(”INSERT INTO “, in_sequenceName, “Sequence (processID) VALUES(?)”);
PREPARE stmt FROM @stmt_text;
SET @a=”1″;
EXECUTE stmt USING @a;
SELECT LAST_INSERT_ID();
END;

It would be called like this:

call pNextVal(’BatchID’);

However, to be even more safe, I guess I can change the sequence table to use InnoDB and do SELECT FOR UPDATE on the stored procedure. Off course it’ll be within transaction.

I just added a few lines before closing the IF to enter the sequence automatically failing:

[...]
WHERE
sequence_name = seq_name
;
INSERT INTO
sequence.sequence_data
SET
sequence_cur_value = 0,
sequence_name = seq_name;
END IF;
[...]