Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
WindChaser
Posting Yak Master
225 Posts |
 Posted - 2006-06-22 : 15:57:59
|
| Hi folks!I'm currently studying the possibility of implementing transactional replication between 2 distant servers, but I'm a little confused about it. I've tried to find a description of the actual synchronization process in BO but with no success. Basically, I want to understand how the process can assure perfect synchronization at transactional level when both servers are logging transactions. IOW, take the following example: let's say that I have a table containing an autoincrementing index and that a transaction A is made on server 1 and, almost simultaneously, a transaction B is made on server 2. Is there some kind of server lock involved where server A locks server B until transaction A is completed on both servers, then the lock is released, then server B locks server A, etc....?Also, can transactional level replication be successfully implemented when the 2 servers are distant using a public ADSL connection instead of a T1/T3 dedicated connection?Thanks all. |
|
|
nathans
Aged Yak Warrior
938 Posts |
Posted - 2006-06-29 : 17:02:19
|
| The only replication scenario that offers no possibility of a conflict is where the subscriber is treated as Read-Only and is NOT updateable. All data changes flow through the publisher.If you need to run updateable subscriptions the very conflict you describe is a possibility. Look into bi-directional transactional replication if these are your requirements.Read more here:[url]http://msdn2.microsoft.com/en-us/library/ms151706.aspx[/url]What is the purpose of the 2nd server? Standby, Reporting, etc. There are many High Availability options to choose from. Replication is only one.Nathan Skerl |
 |
|
|
WindChaser
Posting Yak Master
225 Posts |
Posted - 2006-06-29 : 22:27:36
|
| Thanks for the reply nathans.The client has 2 very medical clinics where there's very active billing, scheduling, etc... continuously. The client wants both sets of DBs to mirror themselves at all times. This means that the DBs at clinic A must be locked before any transaction can occur at clinic B, then mirrored, then released, and vice versa. I was hoping that this was the process behind transactional replication.Thanks. |
|
|
|
nathans
Aged Yak Warrior
938 Posts |
Posted - 2006-06-29 : 22:44:29
|
| Well, it can be. What you want is Bidirectional Transactional Replication. In this scenario each replication node acts as both a publisher and subscriber.Is it even a possiblity that you run a single central server between the two sites?Nathan Skerl |
|
|
|
|
|
|
|
|
Understanding transactional replication