Question

I’m running MySQL 5.1.36 and have a database used for a web-based help desk system. The database has three tables I’d like to track changes for:

CREATE TABLE IF NOT EXISTS `tickets` (
  `TicketNum` int(11) unsigned NOT NULL,
  `SubmittedFromIP` tinyblob,
  `SubmittedFromDevice` varchar(255) DEFAULT NULL,
  `EntryDate` datetime DEFAULT NULL,
  `ClosedDate` datetime DEFAULT NULL,
  `LastName` varchar(50) DEFAULT NULL,
  `FirstName` varchar(50) DEFAULT NULL,
  `Email` varchar(50) DEFAULT NULL,
  `Location` varchar(4) DEFAULT NULL,
  `InventoryNumber` varchar(50) DEFAULT NULL,
  `DeviceName` varchar(50) DEFAULT NULL,
  `Description` text,
  `Notes` text,
  `Agent_ID` smallint(5) unsigned NOT NULL DEFAULT '1',
  `TotalHoursSpent` float NOT NULL DEFAULT '0',
  `Status` smallint(5) unsigned NOT NULL DEFAULT '1',
  `Priority` tinyint(4) NOT NULL DEFAULT '0',
  `LastUpdatedByAgent_ID` smallint(5) unsigned DEFAULT NULL,
  PRIMARY KEY (`TicketNum`),
  KEY `ClosedDate` (`ClosedDate`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `ticketsolutions` (
  `Entry_ID` int(10) unsigned NOT NULL,
  `TicketNum` mediumint(8) unsigned DEFAULT NULL,
  `EntryDateTime` datetime DEFAULT NULL,
  `HoursSpent` float DEFAULT NULL,
  `Agent_ID` smallint(5) unsigned DEFAULT NULL,
  `EntryText` text,
  `LastUpdatedByAgent_ID` smallint(5) unsigned DEFAULT NULL,
  PRIMARY KEY (`Entry_ID`),
  KEY `TicketNum` (`TicketNum`),
  KEY `EntryDateTime` (`EntryDateTime`),
  KEY `HoursSpent` (`HoursSpent`),
  KEY `Rating` (`Rating`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

CREATE TABLE IF NOT EXISTS `tickettagsmap` (
  `TicketNum` int(11) unsigned NOT NULL,
  `Tag_ID` int(10) unsigned NOT NULL,
  `AddedByAgent_ID` smallint(5) unsigned NOT NULL,
  `DateTimeAdded` datetime NOT NULL,
  PRIMARY KEY (`TicketNum`,`Tag_ID`),
  KEY `Tag_ID` (`Tag_ID`),
  KEY `fk_AgentID` (`AddedByAgent_ID`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

Based on what I’ve read, the best way to handle this is to create duplicate tables, only with two extra fields per table:

ModifiedDateTime
Action

Is this really the best way? Every time even the slightest change is made to a record, the entire record is inserted into its corresponding history table. It seems like a huge waste of space. Is there a better way to do this?

Answer 1

It may or may not be a waste of space depends on typical operations with tables. For INSERT and DELETE the only way to track is to store all column values. Only for UPDATE you can save some space. You can create 2 tables, for instance,

update_history_main(id int not null auto_increment primary key,
modify_date datetime not null,
table_involved varchar(50) not null);

update_history_details (id int not null auto_increment primary key,
update_history_main_id int not null,
field_name varchar(100),
old_value varchar(100),
new_value varchar(100),
FOREIGN KEY (update_history_main_id) REFERENCES update_history_main(id)
ON UPDATE CASCADE ON DELETE CASCADE);

and add records into these tables after each update. The problem here is that old_value and new_value columns should be large enough to keep value of any column from your original tables. So you probably need to create another update_history_details_text_blobs that tracks only changes in text/blob columns.

Update. Thus, the body of your after update trigger for tickets table may look like

DELIMITER $$$
 CREATE TRIGGER afterTicketUpdate AFTER UPDATE ON tickets
 FOR EACH ROW
 BEGIN
     DECLARE main_id int;
     INSERT INTO update_history_main(modify_date, table_involved) 
      VALUES(NOW(),'tickets';
     SELECT LAST_INSERT_ID() INTO main_id;

     IF (new.SubmittedFromDevice != old.SubmittedFromDevice) THEN
      INSERT INTO update_history_details(update_history_main_id, field_name, 
      old_value,new_value) 
     VALUES (main_id, 'SubmittedFromDevice',old.SubmittedFromDevice,
        new.SubmittedFromDevice);
     END IF; // ... check all other fields. 
 END
$$$