How to avoid identity RowGUIDcol?
-
Hi,
i am trying to implement merge replication. i have to use identity partitioning using even and odd series on publisher and subscriber (both on different servers). i have only 1 publisher and 1 subscriber.
I have read Micheal Hotek's article of archetecting identities. i have setup a d/b1 with schema for odd identity series and i created another d/b2 on server2 and set the identity columns for even series. on both d/b's i have put the not for replication and populated d/b1 with all the data. Finally i created the distributor on server 1 and when i am trying to configure a publisher for d/b1, i am getting the article issue of ' uniqueidentitfier of type rowcolguid will be added to d/b'. How can i make replication take my identity columns and not let it add rowguid (I thought the entire exercise of changing the d/b schema to accomodate Not For Replication and putting identity seed and increment of 2 was to avoid appending of Rowguid col). the database has about 40 million records, performance is definitely the issue.
Can somebody please help?
thanks
(Also, i have seen one article that says, add a column and set it to uniqueidentifier and rowguidcol property on all the tables. if i follow this how can i implement eve/odd series)
-
By default Merge Replication will place rowguid columns on all Tables that are in a Publication and you cannot stop that because SQL Server will create triggers &/or Stored procedures that the Replication process will use. Don't worry about the RowGuid Columns, they have to be there!
-
The RowGUID is how SQL Server identifies each row and keeps track of changes made to it, if u add them yourself, then u can have control of the name, otherwise, u will have the default SQL server name on each table.
Basically without the RowGUID, replication would not be possible
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply