Publication of mixture without altering the database

[HatRix]

I need to merge several databases that are isolated from each other on different machines that are in different places into one central database and turn it once unified send all data to the machines to have the same information central server. The unification would take place once in a while. I have read in the help of sql server you could do with the publication of mixtures but when I try it tells me I have to add a field in all tables of type foreign. All tables of the databases that attempt to unify with a field (called code) that is itself a unique index that cannot be repeated, but of course this field is in and I generate the values on my application, i.e. it is not, nor foreign Indentity. Also I cannot change the structure of databases as there is an application that works as is and if I change a table gives me errors in the application. How I can then apply merge replication or other or do otherwise without replication without having to alter the structure of databases?

[Balamohan]

Your problem is complex and requires a complex solution. You can opt for a merge replication or transactional, it will depend on a number of factors you should evaluate. I would opt for the compromise, but there may be factors that tip the balance the other site. With transactional replication detail With merge replication detail There you will find detailed information to carry out the implementation. So just go through this.

[Bansi_WADIA]

The initialization will be more complicated, because you have no unified data. I recommend that you take the data which is missing from the central server by other mechanisms, such as Integration Services packages. Required to configure it the same you say you fail (not to exceed calculated fields, overwrite or not the identities, etc). All this to ensure that there is a point that has all the information and from there to initialize the subscription. And it gets import and export of data may help prepare the packages, but you cannot use them as is because the implementation process will require a complicated processing of information. In short, you have a set of packages that will be launched in the proper order to comply with referential integrity. So I know about it only thsmuch. I hope through this you get some idea.

[xanix]

In this the main thing which is to do is pick each database that is distributed on a SQL server manager to copy the data from each into a new database with the same scheme, a unified view your database the public on a central server, then connected every one of these machines to the central server. So just follow this I am sure you will get fix for this issue.

[HatRix]

I have not understood it correctly when you say "copy the data from each in a new database with the same scheme, once unified ", you mean to back up each bd and then restore them in the same host for different DBs and then once in the same instance to publish in a central DB? What type of publication would be made? Compromise, mixing, Instant or what…?