Showing posts with label deploy. Show all posts
Showing posts with label deploy. Show all posts

Monday, March 12, 2012

How best to deploy database?

What would be the best way to deploy a database to customers?

Execute sql scripts to create the database on the customer's server, or distribute an .mdf file?

Package as msi?

Other ideas?

Hi,

if you vuild a database from sctach with only little data shipped with the database, I would go for the scripting. If you have more data you should do something like MSIing. If you don′t want to use an installer just ship the MDF file and leave it to the customer to attach the database.

HTH, jens Suessmeyer.

http://www.sqlserver2005.de|||

Hi Jens,

There is very little data in each newly deployed database. The customer uses an application front end to populate the database to their own specific requirements.

The applicationfront-end will be deployed via MSI as each customer will have multiple installations of it, each pointing at one or more database servers.

Having had little to do with MSI packages thus far is it easy to get an MSI to execute the scripts and/or attach a template MDF?

The way I see it, the database installer will execute under admin account on the server (same as account used to install SQL Server) so can connect using Windows Authentication. At that point I need to execute the setup scripts and/or attach the MDF. Further I need to execute scripts to generate a unique logon used as an application role per SQL Server instance so that the front-end application(s) can connect remotely.

Does this sound a reasonable way to go about it?

Many thanks for your input.

|||

That sounds good, I would offer the user a possibility to choose the instance where he wants to attach the database to (if more than one instance is present). Furtherone I would let him to choose to either connect to the server via SQL Server authentication or Windows authentication (because not all servers might have the local adminstrator group as sysadmins).

With that you should have a minimum of support calls :-) (If any problems occur, don′t forget to implement a logging mechanism which get the output of the executed script for debugging purposes)

HTH, Jens Suessmeyer.

http://www.sqlserver2005.de

Friday, February 24, 2012

Hot Swapping Cluster Nodes

We are trying to deploy a two node Active / Passive cluster. But would like to test / ascertain the necessary steps to hot-swap one of the nodes should the need arise. The scenario being that the node is unavailable for longer than the time deemed accep
table to have a node down.
We would aim to bring a spare node into the cluster, install it wil binaries and allow it to replace the original failover node, whilst the original failover node was being fixed.
However, on initial testing SQL Server seems to have problems with this, and basically, for a node to be part of a cluster it must be present at the initial install.
Is there a way around this?
You can have the node ready to join the cluster but it must actually be in
the cluster before SQL can install to it. Once the node is in the cluster,
you can install SQL and any service packs/hotfixes without taking SQL
offline. BOL has the procedure to replace a failed node in a SQL cluster.
So, you can't do exactly what you want but you can get very close.
Geoff N. Hiten
Microsoft SQL Server MVP
Senior Database Administrator
Careerbuilder.com
I support the Professional Association for SQL Server
www.sqlpass.org
"Beverley" <Bev.Brindle@.barclays.co.uk> wrote in message
news:B41DAA45-424E-4970-ABF9-16243D76CEDC@.microsoft.com...
> We are trying to deploy a two node Active / Passive cluster. But would
like to test / ascertain the necessary steps to hot-swap one of the nodes
should the need arise. The scenario being that the node is unavailable for
longer than the time deemed acceptable to have a node down.
> We would aim to bring a spare node into the cluster, install it wil
binaries and allow it to replace the original failover node, whilst the
original failover node was being fixed.
> However, on initial testing SQL Server seems to have problems with this,
and basically, for a node to be part of a cluster it must be present at the
initial install.
> Is there a way around this?
>
|||Thanks, we're going into a phase of strategic testing for the Cluster, and this is very useful.
Thanks alot.
Bev