Our software is a collection of Windows applications that connect to a SQL database. Currently all our client sites have their own server and SQL Server database, however I'm working on making our software work with Azure-hosted databases too.
I've hit one snag with it, and so far not found anything particularly helpful while Googling around.
The current SQL Server version includes a database auditing system I wrote, which does the following:-
The C# Applications include in the connection string information about which program and version it is, and which User is currently logged in.
Important tables have Update and Delete triggers, which send details of any changes to a Service Broker queue. (I don't log Inserts).
The Service Broker then goes through the queue, and records details of the change to a separate AuditLog table.
These details include:-
Table, PK of the row changed, Field, Old Value, New Value, whether it was an Update or Delete, date/time of change, UserID of the user logged in to our software, and which program and version made the change.
This all works very nicely, and I was hoping to keep the system as-is for the Azure version, but unfortunately SQL Azure does not have Service Broker.
So, I need to look for an alternative, which as I mentioned is proving troublesome.
There is SQL Azure Managed Instances, which does have Service Broker, however they are way too expensive for us to even consider. Not one of our clients would pay that much per month.
Anything else I've looked at doesn't seem to have everything I need. In particular, logging which program, version and UserID. Note that this isn't the SQL login UserID, which will be the same for everyone, this is the ID from the Users table with which they log in to our software, and is passed in the Connection String.
So, ideally I'd like something similar to what I have, just with something else in the place of the Service Broker:-
The C# Applications include in the connection string information about which program and version it is, and which User is currently logged in.
Important tables have Update and Delete triggers, which send details of any changes to an asynchronous queue of some sort.
Something then goes through the queue outside the normal program flow, and records details of the change to a separate AuditLog table.
The asynchronous queue and processing outside the normal program flow is important. Obviously I could very easily have the Update and Delete triggers do all the processing and add the records to the AuditLog table, in fact that was v1.0 of the system, but the problem there is that SQL will wait until the triggers have finished before returning to the C# program. This then causes the C# program to slow down considerably when multiple Updates or Deletes are happening.
I'd be happy to look into other logging systems instead of the above, however something which only records data changes without the extra information I pass, specifically program, version and UserID, won't be of any use to me. Our Users always want to know this information whenever they query something they think is an incorrect change.
So, any suggestions for an alternative to Service Broker for SQL Azure please? TIA!
Ok, looks like I have a potential solution: Temporal Tables
Temporal Tables work in Azure, and record a new row in a History table whenever something changes:-
CREATE TABLE dbo.LMSTemporalTest
(
[EmployeeID] INT NOT NULL PRIMARY KEY CLUSTERED
, [Name] NVARCHAR(100) NOT NULL
, [Position] NVARCHAR(100) NOT NULL
, [Department] NVARCHAR(100) NOT NULL
, [Address] NVARCHAR(1024) NOT NULL
, [AnnualSalary] DECIMAL (10,2) NOT NULL
, [UpdatedBy] UniqueIdentifier NOT NULL
, [UpdatedDate] DateTime NOT NULL
, [ValidFrom] DateTime2 (2) GENERATED ALWAYS AS ROW START HIDDEN
, [ValidTo] DateTime2 (2) GENERATED ALWAYS AS ROW END HIDDEN
, PERIOD FOR SYSTEM_TIME (ValidFrom, ValidTo)
)
WITH (SYSTEM_VERSIONING = ON (HISTORY_TABLE = dbo.LMSTemporalTestHistory));
GO
I can then insert a record into the table...
INSERT INTO LMSTemporalTest(EmployeeID,Name,Position,Department,Address,AnnualSalary, UpdatedBy, UpdatedDate)
VALUES(1, 'Bob', 'Builder', 'Fixers','Oops I forgot', 1, '0D7F5584-C79B-4044-87BD-034A770C4985', GetDate())
GO
Update the row...
UPDATE LMSTemporalTest SET
Address = 'Sunflower Valley, Bobsville',
UpdatedBy = '2C62290B-61A9-4B75-AACF-02B7A5EBFB80',
UpdatedDate = GetDate()
WHERE EmployeeID = 1
GO
Update the row again...
UPDATE LMSTemporalTest SET
AnnualSalary = 420.69,
UpdatedBy = '47F25135-35ED-4855-8050-046CD73E5A7D',
UpdatedDate = GetDate()WHERE EmployeeID = 1
GO
And then check the results:-
SELECT * FROM LMSTemporalTest
GO
EmployeeID Name Position Department Address AnnualSalary UpdatedBy UpdatedDate
1 Bob Builder Fixers Sunflower Valley, Bobsville 420.69 47F25135-35ED-4855-8050-046CD73E5A7D 2019-07-01 16:20:00.230
Note: Because I set them as Hidden, the Valid From and Valid To don't show up
Check the changes for a date / time range:-
SELECT * FROM LMSTemporalTest
FOR SYSTEM_TIME BETWEEN '2019-Jul-01 14:00' AND '2019-Jul-01 17:10'
WHERE EmployeeID = 1
ORDER BY ValidFrom;
GO
EmployeeID Name Position Department Address AnnualSalary UpdatedBy UpdatedDate
1 Bob Builder Fixers Oops I forgot 1.00 0D7F5584-C79B-4044-87BD-034A770C4985 2019-07-01 16:20:00.163
1 Bob Builder Fixers Sunflower Valley, Bobsville 1.00 2C62290B-61A9-4B75-AACF-02B7A5EBFB80 2019-07-01 16:20:00.197
1 Bob Builder Fixers Sunflower Valley, Bobsville 420.69 47F25135-35ED-4855-8050-046CD73E5A7D 2019-07-01 16:20:00.230
And I can even view the History table
SELECT * FROM LMSTemporalTestHistory
GO
EmployeeID Name Position Department Address AnnualSalary UpdatedBy UpdatedDate ValidFrom ValidTo
1 Bob Builder Fixers Oops I forgot 1.00 0D7F5584-C79B-4044-87BD-034A770C4985 2019-07-01 16:20:00.163 2019-07-01 16:20:00.16 2019-07-01 16:20:00.19
1 Bob Builder Fixers Sunflower Valley, Bobsville 1.00 2C62290B-61A9-4B75-AACF-02B7A5EBFB80 2019-07-01 16:20:00.197 2019-07-01 16:20:00.19 2019-07-01 16:20:00.22
Note: the current row doesn't show up, as it's still Valid
All of our important tables have CreatedBy, CreatedDate, UpdatedBy and UpdatedDate already, so I can use those for the UserID logging. No obvious way of handling the Program and Version as standard, but I can always add another hidden field and use Triggers to set that.
EDIT: Actually tested it out
First hurdle was: can you actually change an existing table into a Temporal Table, and the answer was: yes!
ALTER TABLE Clients ADD
[ValidFrom] DateTime2 (2) GENERATED ALWAYS AS ROW START HIDDEN NOT NULL DEFAULT '1753-01-01 00:00:00.000',
[ValidTo] DateTime2 (2) GENERATED ALWAYS AS ROW END HIDDEN NOT NULL DEFAULT '9999-12-31 23:59:59.997',
PERIOD FOR SYSTEM_TIME (ValidFrom, ValidTo)
GO
ALTER TABLE Clients SET (SYSTEM_VERSIONING = ON (HISTORY_TABLE = dbo.ClientsHistory))
GO
An important bit above is the defaults on the ValidFrom and ValidTo fields. It only works if ValidTo is the maximum value a DateTime2 can be, hence '9999-12-31 23:59:59.997'. ValidFrom doesn't seem to matter, so I set that to the minimum just to cover everything.
Ok, so I've converted a table, but it now has two extra fields that the non-Azure table doesn't, which are theoretically hidden, but will our software complain about them?
Seems not. Fired up the software, edited a record on the Clients table and saved it, and the software didn't complain at all.
Checked the Clients and ClientsHistory tables:-
SELECT * FROM Clients
FOR SYSTEM_TIME BETWEEN '1753-01-01 00:00:00.000' AND '9999-12-31 23:59:59.997'
WHERE sCAccountNo = '0001064'
ORDER BY ValidFrom
Shows two records, the original and the edited one, and the existing UpdatedUser and UpdatedDate fields show correctly so I know who made the change and when.
SELECT * FROM ClientsHistory
Shows the original record, with ValidTo set to the date of the change,
All seems good, now I just need to check that it still only returns the current record in queries and to our software:-
SELECT * FROM Clients
WHERE sCAccountNo = '0001064'
Just returned the one record, and doesn't show the HIDDEN fields, ValidFrom and ValidTo.
Did a search in our software for Client 0001064, and again it just returned the one record, and didn't complain about the two extra fields.
Still need to set up a few Triggers and add another HIDDEN field to record the program and version from the Connection String, but it looks like Temporal Tables gives me a viable audit option.
Only downside so far is that it creates an entire record row for each set of changes, meaning you have to compare it to other records to find out what changed, but I can write something to simplify that easily enough.
Related
I recently migrated my SQL 2019 database from a VM into Azure SQL.
I used the MS Data Migration tool, but unfortunately, it wouldn't migrate data from Temporal Tables.
So. I just used the tool to create the table schemas and then used SSIS to move the data.
Since my existing history table had data in it, I wanted to keep the SysStartDate and SysEndDate fields. In order to do this, I had to disable SYSTEM_VERSIONING in my Azure SQL database as well as DROP the PERIOD on the table.
The data migration was a success so I re-created my PERIOD on the table but when I tried to enable SYSTEM_VERSIONING with a specified history table, I get the following error:
Msg 13573, Level 16, State 0, Line 34
Setting SYSTEM_VERSIONING to ON failed because history table 'xxxxxHistory' contains overlapping records.
I find this odd because the existing tables were originally joined as a temporal table so I don't understand why there would be a conflict now.
ALTER TABLE xxx.xxx
ADD PERIOD FOR SYSTEM_TIME(SysStartTime, SysEndTime)
ALTER TABLE xxx.xxx
SET (SYSTEM_VERSIONING = ON (HISTORY_TABLE=xxx.xxxHistory))
I expect to get a successful temporal table. Instead, I get the following error:
Msg 13573, Level 16, State 0, Line 34
Setting SYSTEM_VERSIONING to ON failed because history table 'xxxxxHistory' contains overlapping records.
I ran the following query to identify the overlaps but I don't get any:
SELECT
xxxxKeyNumeric
,SysStartTime
,SysEndTime
FROM
xxxx.xxxxhistory o
WHERE EXISTS
(
SELECT
1
FROM
xxxx.xxxxhistory o2
WHERE
o2.xxxxKeyNumeric = o.xxxxKeyNumeric
AND o2.SysStartTime <= o.SysEndTime
AND o.SysStartTime <= o2.SysEndTime
AND o2.xxxxPK != o.xxxxPK
)
ORDER BY
o.xxxxKeyNumeric,
o.SysStartTime
I found this explanation for the error:
"There are multiple records for the same record with overlapping start and end dates. The end date for the last row in the history table should match the start date for the active record in the parent table" blog of a DBA
This happened to me after switching the historic table, touching a few rows, then trying to go back to the old historic table.
UPDATE: Happened again, and this time the table had millions of rows. I had to write a query, comparing the start date and end date of every row in the history table.
Possible causes:
For every PK, the start dates and end dates of the history rows must not overlap. The query below will find this specific issue.
the end date of the latest row in the history for that PK, has a later end date than the start date of the PK in the main table. It is possible to modify the above query to do this
in the rows with a same PK, 2 rows cover the same time interval. If they overlap by a single millisecond, and someone requests that exact millisecond, it won't know which of the 2 versions is the correct one.
For the first issue:
select ant.*,post.* , DATEDIFF(day,ant.end_date,post.start_date)
from
(SELECT
PK_column
, start_date
, end_date
, ROW_NUMBER() OVER(PARTITION BY PK_column ORDER BY end_date desc, start_date desc) AS current
,(ROW_NUMBER() OVER(PARTITION BY PK_column ORDER BY end_date desc, start_date desc))-1 AS previous
FROM huge_table_HIST
) ant
inner join
(SELECT
key_column
, start_date
, end_date
, ROW_NUMBER() OVER(PARTITION BY PK_column ORDER BY end_date desc, start_date desc ) AS current
FROM huge_table_HIST
) post
ON ant.PK_column=post.PK_column AND ant.previous=post.current
WHERE ant.end_date > post.start_date
Surprisingly, it doesn't fail if:
you have multiple rows with exactly the same start end and end date, for the same PK. SQL Server seems to consider them a single point in space, instead of an interval. They will only appear if you request the exact millisecond in which they exist.
there are gaps between the end date of a history row, and the start end of the next one. SQL server considers that the PK just didn't exist in that time interval.
Temporal tables depend on the temporal table's primary key values combined with the SysStartTime do determine uniqueness in the history table.
This can very easily happen if you make changes to the primary key definition. Also, if your history table's fields corresponding to the temporal table's PK are not populated, or many / all are populated with a default value, overlaps are detected and you get that error.
Check that your PK is defined on the system versioned temporal table, then check that the corresponding values in your history table's primary key fields are correct (i.e. unique for any given PK & SysStartTime value.)
You may have to update the history table accordingly before applying the system versioning relationship again.
This error can also occur when there are multiple records per Primary Key for any given
GENERATED ALWAYS AS ROW START or GENERATED ALWAYS AS ROW END columns.
The following queries will help identify those records.
select ID
from dbo.HistoryTable
group by ID, SysStartTime
having count(*) > 1
select ID
from dbo.HistoryTable
group by ID, SysEndTime
having count(*) > 1
I need to create a service but i need a help with choice of tools.
Imagine service in which users create some data that have value in historical view (e.g. transactions). Other users can see this data but they need a proof that data are real and not falsified by users or even by service.
Example:
User A creates record with number 42
Couple of months passes
User B see this record and wants to be sure that service can't update this record with any other number 37
Service has trust window with 24 hours: it even can change users data, which were made on this day.
Question: Which instruments can help me to achieve that?
I was thinking about doing public daily backups (or reports?) that any user can download. From each report hash will be calculated and inserted into next backup – thus, a chan of hashes created. If service will change something in past, then hashes in this chain will not converge. Of course, i'll create open sourced tool for easy comparing diff between data and check if chain is valid.
Point of trust: there is one thing that i'm afraid of. Service can use many databases simultaneously and update all backups with all hashes one time (because first backup has no hash of previous one). So, to cover that case too, i think of storing hashes in some place that service can't change at all. For example, in one of the existed blockchains (btc, eth, ...) from official wallet of service. Or, maybe, DAG with some blockchain like IOTA?
What do you think of point of trust?
Can i achieve my goal with some simpler way (without blockchain)? And which one?
What are bottlenecks in this logic?
There are 2 participating variables here
timestamp at which the record is created.
the data.
Solution premise,
Tampering proof.
the data can be changed in the same GMT calendar day without violating tamper-proof guarantee. (can be changed to a fixed window after creation)
RDBMS as the data store, (can be changed to any NoSQL with minor mods, but the idea remains the same).
Doesn't depend on any other mechanism which can be faulty or error-prone.
Single query verification.
## Proposed solution
create data table
CREATE TABLE TEST(
ID INT PRIMARY KEY AUTO_INCREMENT,
DATA VARCHAR(64) NOT NULL,
CREATED_AT DATETIME DEFAULT CURRENT_TIMESTAMP()
);
create checksum table, which monitor tempering
CREATE TABLE SIGN(
ID INT PRIMARY KEY AUTO_INCREMENT,
DATA_ID INT NOT NULL,
SIGNATURE VARCHAR(128) NOT NULL,
CREATED_AT DATETIME DEFAULT CURRENT_TIMESTAMP(),
UPDATED_AT TIMESTAMP
);
create trigger on insert of data
/** Trigger on insert */
DELIMITER //
CREATE TRIGGER sign_after_insert
AFTER INSERT
ON TEST FOR EACH ROW
BEGIN
-- INSERT VAL
INSERT INTO SIGN(DATA_ID, `SIGNATURE`) VALUES(
NEW.ID, MD5(CONCAT (NEW.DATA, DATE(NEW.CREATED_AT)))
);
END; //
DELIMITER ;
Create a trigger for update of data
-- UPDATE TRIGGER
DELIMITER //
CREATE TRIGGER SIGN_AFTER_UPDATE
AFTER UPDATE
ON TEST FOR EACH ROW
BEGIN
-- UPDATE VALS
IF (NEW.DATA <> OLD.DATA) AND (DATE(OLD.CREATED_AT) = CURRENT_DATE() ) THEN
UPDATE SIGN SET SIGNATURE=MD5(CONCAT(NEW.DATA, DATE(NEW.CREATED_AT))) WHERE DATA_ID=OLD.ID;
END IF;
END; //
DELIMITER ;
Test
Step 1: insert the data
INSERT INTO TEST(DATA) VALUES ('DATA2');
The signature of data and the date at which it was created, will reflect as the signature in SIGN table.
Step 2: update the data
the signature will get updated if value is changed and it is the SAME DAY.
UPDATE TEST SET DATA='DATA' WHERE ID =1;
Step 3: validate
you can always validate the data signature as
SELECT MD5(CONCAT (T.DATA, DATE(T.`CREATED_AT`))) AS CHECKSUM, S.SIGNATURE FROM TEST AS T ,SIGN AS S WHERE S.DATA_ID= T.ID AND S.`id`=1;
Output
| CHECKSUM | SIGNATURE |
| ------ | ------ |
|2bba70178abdafc5915ba0b5061597fa |2bba70178abdafc5915ba0b5061597fa
I'm working on a leave software, and my problem is that i need to reset the leave days to default number of days (30 day) after one year. would you pleas help me with that.
ps: I'm using VB.NET AND SQL SERVER.
create table Addemployees
(
Fname varchar (500),
Lname varchar (500),
ID int not null identity(1, 1) primary key,
CIN varchar (500),
fromD date,
toD date,
Email varchar(500),
phone varchar(500),
Leave_num int
)
This is the tablet that contains the column Leave_num that has the leave numbers inserted by the user
update addemployees
set leave_num = 30
As for how you trigger this logic. There are many ways you could go about this. You'll need some sort of scheduler like an Agent job, or whatever else you have at your disposal to run this process on a recurring, scheduled, basis. The key thing is not to keep updating the LeaveNum if it's already been updated. You could maintain an extra column on each row indicating the last time they were reset. This is probably the simplest, but if it's truly an all-or-nothing type thing, and those dates will all be the same, that's sort of a waste of space.
You could then either create a separate table which just contains information about when these once-a-year jobs run, or something like an Extended Property (which is a little more involved to set up).
Whatever the solution you choose, Just save off the date (or even just the year), and then when your process runs, if the difference between the last update is greater than a year (or if the year of the last update is less than the current year) run your update, then update however you're storing that information; be it columns, a separate table, or an extended property.
I am using SQL Server 2012
I have a table called AMOUNTS and a table called AMOUNTS_HIST
Both tables have identical columns:
CHANGE_DATE
AMOUNT
COMPANY_ID
EXP_ID
SPOT
UPDATE_DATE [system date]
The Primary Key of AMOUNTS is COMPANY_ID and EXP_ID.
The Primary Key pf AMOUNTS_HIST is COMPANY_ID, EXP_ID and CHANGE_DATE
Whenever I add a row in the AMOUNTS table, I would like to create a copy of it in the AMOUNTS_HIST table. [Theoretically, each time a row is added to 'AMOUNTS', the COMPANY_ID, EXP_ID, CHANGE_DATE will be unique. Practically, if they are not, the relevant row in AMOUNTS_HIST would need to be overridden. The code below does not take the overriding into account.]
I created a trigger as follows:
CREATE TRIGGER [MYDB].[update_history] ON [MYDB].[AMOUNTS]
FOR UPDATE
AS
INSERT MYDB.AMOUNTS_HIST (
CHANGE_DATE,
COMPANY_ID,
EXP_ID,
SPOT
UPDATE_DATE
)
SELECT e.CHANGE_DATE,
e.COMPANY_ID,
e.EXP_ID
e.REMARKS,
e.SPOT,
e.UPDATE_DATE
FROM MYDB.AMOUNTS e
JOIN inserted ON inserted.company_id = e.company_id
AND inserted.exp_id=e.exp_id
I don't understand why it does nothing at all in my AMOUNTS_HIST table.
Can anyone help?
Thanks,
Probably because the trigger, the way it's currently written, will only get fired when an Update is done, not an insert.
Try changing it to:
CREATE TRIGGER [MYDB].[update_history] ON [MYDB].[AMOUNTS]
FOR UPDATE, INSERT
I just wanted to chime in. Have you looked at CDC (change data capture).
http://msdn.microsoft.com/en-us/library/bb522489(v=sql.105).aspx
"Change data capture is designed to capture insert, update, and delete activity applied to SQL Server tables, and to make the details of the changes available in an easily consumed relational format. The change tables used by change data capture contain columns that mirror the column structure of a tracked source table, along with the metadata needed to understand the changes that have occurred.
Change data capture is available only on the Enterprise, Developer, and Evaluation editions of SQL Server."
As far as your trigger goes, when you update [MYDB].[AMOUNTS] does the trigger throw any errors?
Also I believe you can get all your data from Inserted table without needed to do the join back to mydb.amounts.
My original question can be found here, for which I've gotten some great answers, idas and tips.
As part of a feasibility and performance study, I've started to convert my schemas in order to version my data using those ideas. In doing so, I've come up with some kind of other problem.
In my original question, my example was simple, with no real relational references. In an attempt to preserve the example of my previous question, I will now extend the 'Name' part to another table.
So now, my data becomes:
Person
------------------------------------------------
ID UINT NOT NULL,
NameID UINT NOT NULL,
DOB DATE NOT NULL,
Email VARCHAR(100) NOT NULL
PersonAudit
------------------------------------------------
ID UINT NOT NULL,
NameID UINT NOT NULL,
DOB DATE NOT NULL,
Email VARCHAR(100) NOT NULL,
UserID UINT NOT NULL, -- Who
PersonID UINT NOT NULL, -- What
AffectedOn DATE NOT NULL, -- When
Comment VARCHAR(500) NOT NULL -- Why
Name
------------------------------------------------
ID UINT NOT NULL,
FirstName VARCHAR(200) NOT NULL,
LastName VARCHAR(200) NOT NULL,
NickName VARCHAR(200) NOT NULL
NameAudit
------------------------------------------------
ID UINT NOT NULL,
FirstName VARCHAR(200) NOT NULL,
LastName VARCHAR(200) NOT NULL,
NickName VARCHAR(200) NOT NULL,
UserID UINT NOT NULL, -- Who
NameID UINT NOT NULL, -- What
AffectedOn DATE NOT NULL, -- When
Comment VARCHAR(500) NOT NULL -- Why
In a GUI, we could see the following form:
ID : 89213483
First Name : Firsty
Last Name : Lasty
Nick Name : Nicky
Date of Birth : January 20th, 2005
Email Address : my.email#host.com
A change can be made to:
Only to the 'name' part
Only to the 'person' part
To both the 'name' and person parts
If '1' occurs, we copy the original record to NameAudit and update our Name record with the changes. Since the person reference to the name is still the same, no changes to Person or PersonAudit are required.
If '2' occurs, we copy the original record to PersonAudit and update the Person record with the changes. Since the name part has not changed, no changes to Name or NameAudit are required.
If '3' occurs, we update our database according to the two methods above.
If we were to make 100 changes to both the person and name parts, one problem occurs when you later try to show a history of changes. All my changes show the person having the last version of the name. Which is wrong obviously.
In order to fix this, it would seem that the NameID field in Person should reference the NameAudit instead (but only if Name has changes).
And it is this conditional logic that starts complicating things.
I would be curious to find out if anyone has had this kind of problem before with their database and what kind of solution was applied?
You should probably try to read about 'Temporal Database' handling. Two books you could look at are Darwen, Date and Lorentzos "Temporal Data and the Relational Model" and (at a radically different extreme) "Developing Time-Oriented Database Applications in SQL", Richard T. Snodgrass, Morgan Kaufmann Publishers, Inc., San Francisco, July, 1999, 504+xxiii pages, ISBN 1-55860-436-7. That is out of print but available as PDF on his web site at cs.arizona.edu. You can also look for "Allen's Relations" for intervals - they may be helpful to you.
I assume that the DATE type in your database includes time (so you probably use Oracle). The SQL standard type would probably be TIMESTAMP with some number of fractional digits for sub-second resolution. If your DBMS does not include time with DATE, then you face a difficult problem deciding how to handle multiple changes in a single day.
What you need to show, presumably, is a history of the changes in either table, with the corresponding values from the other table that were in force at the time when the changes were made. You also need to decide whether what you are showing is the before or after image; presumably, again, the after image. That means that you will have a 'sequenced' query (Snodgrass's term), with columns like:
Start time -- When this set of values became valid
End time -- When this set of values became invalid
PersonID -- Person.ID (or PersonAudit.ID) for Person data
NameID -- Name.ID (or NameAudit.ID) for Name data
DOB -- Date of Birth recorded while data was valid
Email -- Email address recorded while data was valid
FirstName -- FirstName recorded while data was valid
LastName -- LastName recorded while data was valid
NickName -- NickName recorded while data was valid
I assume that once a Person.ID is established, it does not change; ditto for Name.ID. That means that they remain valid while the records do.
One of the hard parts in this is establishing the correct set of 'start time' and 'end time' values, since transitions could occur in either table (or both). I'm not even sure, at the moment, that you have all the data you need. When a new record is inserted, you don't capture the time it becomes valid (there is nothing in the XYZAudit table when you insert a new record, is there?).
There's a lot more could be said. Before going further, though, I'd rather have some feedback about some of the issues raised so far.
Some other SO questions that might help:
Data structure for non-overlapping ranges within a single dimension
Why do we need a temporal database
Determine whether two date ranges overlap
Best relational database representation of time-bound hierarchies
Since this answer was first written, there's another book published about another set of methods called 'Asserted Versioning' for handling temporal data. The book is 'Managing Time in Relational Databases: How to Design, Update and Query Temporal Data' by Tom Johnston and Randall Weiss. You can find their company at AssertedVersioning.com. Beware: there may be patent issues around the mechanism.
Also, the SQL 2011 standard (ISO/IEC 9075:2011, in a number of parts) has been published. It includes some temporal data support. You can find out more about that and other issues related to temporal data at TemporalData.com, which is more of a general information site rather than one with a particular product axe to grind.
Keep a single changes table with an autoincrement ID and make all your changes to refer to that table.
Always put the original record to the audit table.
To build a history, select all your changes and show the value closest to the change.
Like this:
`Change`
1
2
3
4
5
6
`NameAudit`
1 - created as John Smith
5 - changed to James Smith
`PersonAudit`
1 - created as born on `01.01.1980` in `Seattle, WA`
2 - changed DOB to '01.01.1980`
3 - changed DOB to '02.01.1980`
4 - changed DOB to '02.01.1980`
6 - changes POB to `Washington, DC`
Then select:
SELECT c.id,
(
SELECT MAX(id)
FROM NameAudit na
WHER na.id <= c.id
) as nameVersion,
(
SELECT MAX(id)
FROM PersonAudit pa
WHER pa.id <= c.id
) as personVersion,
na.*,
pa.*
FROM change c
JOIN NameAudit na
ON na.id = nameVersion
JOIN PersonAudit pa
ON pa.id = nameVersion
WHERE change_id BETWEEN 1 AND 6