Database Modelling in UML

The Methods & Tools newsletter has released in its archive section the
article "Database Modelling in UML". This article offers a step by step
approach to transform a UML object-oriented class model in a purely
relational database.

http://www.methodsandtools.com/archive/archive.php?id=9Franco Martinig via SQLMonster.com wrote:

> The Methods & Tools newsletter has released in its archive section the
> article "Database Modelling in UML". This article offers a step by step
> approach to transform a UML object-oriented class model in a purely
> relational database.
> http://www.methodsandtools.com/archive/archive.php?id=9

The article is rather difficult to believe after lines like this:

"The relational model only offers a public data access model. All data is
equally exposed and open to any process to update, query or manipulate it.
Information hiding is unknown."

This statement is true in the same sense that it is true that you have to
turn a crank on the front of a car to get it started. It depends on
whether or not you are Rip Van Winkle.

I kept reading looking for a qualifier, but the mistake is repeated later
on, "The relational model exposes all data equally, with limited support
for associating behaviour with data elements through triggers, indexes and
constraints."

--
Kenneth Downs
Secure Data Software, Inc.
(Ken)nneth@.(Sec)ure(Dat)a(.com)|||Perhaps for another perspective, read:
http://www.dbdebunk.com/page/page/622916.htm
http://www.dbdebunk.com/page/page/622867.htm

--
Anith

database mirroring, methods for keeping logins synced.

I am in the process of implementing database mirroring. I am trying to get an understanding about manageable ways of keeping sql user logins synced between instances.

I have looked at KB918992 which makes sense for the initial intallation, but what about keeping passwords synced in the long-term. I don't want to have to manually set a users password on the failover every time they change it.

How is everyone else accomplishing this using sql logins (not windows auth)? Pointers to articles and/or documentation would be great.

Thanks.

Hi,

Came across this while trying to find out how to ensure that SQL logins have the same SIDs on the mirror server as well as the prinicipal server. There's a script shown that details how to create all the logins on the mirror server from the principal server. You could just add something to drop them at the start and run it every night or whenever necessary.

Hope this helps

http://www.mssqltips.com/tip.asp?tip1166

Database Mirroring, client connections all ODBC so what methods are recommended?

As stated in the subject I have a situation where if database mirroring is employed for either manual or automatic failover, all the client (including web connections) connections use ODBC not ADO, or OLEDB etc... so what methods are recommended? Client side redirect is not available so I could not employe the "Data Source =A; Failover Partner=B..." option.

Right now the method employed (pre database mirroring and basically employing log shipping on SQL 2000) is to have a DNS alias for the ODBC connection so that if the server were to change in a failover situation the DNS record would have to be altered, so that all the client connections would not have to be reconfigured.

Regards,

Dominic Baines

I have one customer in production that uses the DNS name change method to switch over the clients after a failover, and so that method does have uses.

Alternatively, you could have the application handle the failover by issuing the redirect to another server.

Thanks,

Mark

|||

Sorry Mark but just not an option to recode app connection strings.

Several reasons:

1. Most apps are 3rd party and insist on ODBC only connections (I know old method but that is state of affairs).

2. Recoding ODBC connections means we would have to visit physically or by logon/registry script every client to reconfigure connection when failover occurs. Not practical and bound to miss some.

3. When fail back occurs same problem exists in reverse.

We simply have too many applications and over 50 DB's on 6 pairs of servers in the failover/failback situation (canot use clustering) and 100+ app servers and many many direct connect clients involved. And the desktop clients spread over 5 countries including several timezones and as most everyone else these days also India.

It looks like DNS alias is the option unless anyone has a better one? It was originally coded for older NT4 style domain then migrated to AD, which was interesting but that is a different story.

As this appears to be the case can the failover be used to trigger a script in SQL2005 (that will update the DNS (I can handle that script). How could you detect the failover has occured? Still a bit too new to SQL2005 to know the answer to this one.

(and...thanks for the message on the webcast, Kimberley has got in touch).

Regards,

Dom

|||

I understood that the SQL 2005 ODBC provider (SQL Native Client ODBC provider) was part of the SNAC stack and therefore would support failover. Of course the SQL 2005 SNAC bits and pieces would need to be installed on the clients/servers and the connection string(s)/DSNs modified to support a mirror, but that would be a one off operation, albeit maybe a large one.

As per BOL : "The SQL Native Client ODBC driver complies with the Microsoft Win32? ODBC 3.51 specification. The driver supports applications written with the ODBC 2.5 or earlier versions of the ODBC functions in the manner defined in the ODBC 3.51 specification."

Am I missing something?

|||

I have marked as answer as basically the solution does provide an answer that will probably assist some.

However, I will carry on using the DNS method for the moment.

There is no scope right now to upgrade the native ODBC drivers as I am 'stuck' with desktops with older drivers and reconfiguring the client driver or connections is not possible as it would also mean rolling out new desktop images.

Regards,

Dom

sql