How to set up SQL Server Transactional Replication : Multiple Publishers, Single Subscriber


SQL ReplicationWhile providing a replication solution to one of my client in SQL Server 2016, I came across a scenario where the client had retail stores at various locations and all data from those stores had to be synchronized at a central location in real time. The aim of this synchronization was to make it easy for senior management to analyze all stores performance from a single location. After giving considerations to different options, I finally came up with a solution that worked for my client.


Let’s assume we have server  RBTSERV1 and RBTSERV2, which will act as publisher for 2 stores. RBTSERV3 will act as a subscriber and data from both servers RBTSERV1 and RBTSERV2 will be synchronized to RBTSERV3.


There were several tables that needed to be synchronized but to keep it simple, I will take 1 table in our example.


Notice that All stores have a unique StoreId assigned to them.Let’s create some sample table and data on all servers.



create database DBPub1

use DBPub1

create table CustomerOrders(StoreID int ,OrderId int, ItemName varchar(100), Qty int, OrderDate datetime default getdate()

, CONSTRAINT PK_CustomerOrders PRIMARY KEY (StoreID,OrderId))


INSERT INTO CustomerOrders VALUES (1,1,’Books’,5,Getdate())

INSERT INTO CustomerOrders VALUES (1,2,’Toys’,3,Getdate())



create database DBPub2

use DBPub2

create table CustomerOrders(StoreID int ,OrderId int, ItemName varchar(100), Qty int, OrderDate datetime default getdate()

, CONSTRAINT PK_CustomerOrders PRIMARY KEY (StoreID,OrderId))


INSERT INTO CustomerOrders VALUES (2,1,’fans’,15,Getdate())

INSERT INTO CustomerOrders VALUES (2,2,’pens’,13,Getdate())



create table CustomerOrders(StoreID int ,OrderId int, ItemName varchar(100), Qty int, OrderDate datetime default getdate()

, CONSTRAINT PK_CustomerOrders PRIMARY KEY (StoreID,OrderId))


So publisher RBTSERV1 has 2 records in the table and publisher RBTSERV2 also has 2 records in the table. Subscriber doesn’t have any records.


Now let’s start with our replication.


Configuring Publication on RBTSERV1:

To setup a publication on RBTSERV1, Follow the below steps:

  1. Right Click on folder “Local Publication” and select “New Publication…”



  1. In the publication wizard, select the database to be used for Publication. In our case, we are using DBPub1.



  1. Since we have to make the sync almost real time, so we choose Transaction replication.




  1. Next we pick the table we wish to use for our replication. Here the important point is to choose the “Properties for all Table Articles” under “Article Properties” as below:

Action if name is in use: Keep Existing Object Unchaged





  1. Next we leave the “Create a Snapshot Immediately…” option unchecked.



  1. Under Agent security, choose the options below. Note that on my server, SQL and Agent services are running under administrator account.



  1. Give your publication a name and click Finish.


  1. The publication wizard should finish without any error.



  1. Next repeat the same steps for RBTSERV2 and give your publication a name SERV2_Publication.
  2. Now right click on Publication SERV1_Publication and choose “New Subscriptions…”.  In the Subscription wizard, choose your publication as SERV1_Publication





  1. Then you select if you wish to have a push or pull subscription. Here I have used Push Subscriptions






  1. Now you select your subscriber name and the subscription database. Here I have used RBTSERV3 as subscriber and DBSubs as subscriber database.



  1. Configure the distributor Agent security.



  1. Choose to run the distributor Agent Continuously.



  1. It is important not to initialize the subscription as we will be manually synchronizing the old data.




  1. Finally finish the subscription wizard.



  1. Now if you check the log reader agent status of RBTSERV1 or RBTSERV2, you should see it running fine.



  1. And now if you check data in subscriber table using below query, there should be no data.SELECT *   FROM [DBSubs].[dbo].[CustomerOrders]


Now let’s add some new records to both servers.



INSERT INTO CustomerOrders VALUES (1,5,’notebooks’,15,Getdate())


INSERT INTO CustomerOrders VALUES (2,6,’erasers’,150,Getdate())

  1. Now finally check that the new records from both stores have come to subscriber table.



  1. Now you can sync the old records from both stores either using a SQL query or export import wizard. Now in future if you wish to reinitialize just one store, you may delete the data of that store from subscriber and manually sync the data from that store using queries

or export import wizard.

So that’s that. Hope it helped. If you have any issue in understanding the article, feel free to contact us and we will surely help you. Call or whatsapp – +91 997148322


email- support@redbushtechnologies.


Author- Suresh Kumar – A seasoned SQL DBA with more than 15 years of experience in working with fortune 500 MNCs.


Thank you

Team RedBush





Please follow and like us:
This entry was posted in SQL DBA Tutorials and tagged , , , , , , . Bookmark the permalink.

4 Responses to How to set up SQL Server Transactional Replication : Multiple Publishers, Single Subscriber

  1. AK Moahnty says:

    A very nicely and detailed explained article. What I could not understand is, why did you choose to sync old records later, (after the replication was over)? Any specific reason? Could you also post an article regarding step 20 please.

    Thanks again for a nice article.

    • admin says:

      Old records are synced later because we didn’t use snapshot to initialize the replication. So there is no way for old records to get synchronized. Thats why we have to sync them later. If you use snapshot, then there is no such need as snapshot will sync the old data as well. but in this particular case, we couldn’t use snapshot, so the sync has to be done later using normal SQL Server scripts.

  2. Jyothi says:

    i dont understand one thing, without snapshot agent the above replication implementation how it worked?
    According to your step5…
    I feel,It will give error.. Pls recheck and confirm…

    • admin says:

      Hi Jyoti
      You really don’t need a snapshot always to initialize a replication. So here we are basically starting the replication from the moment it is configured. Old data is not synced as I didn’t use snapshot. So I have to sync it later.
      FYI, replication can also be initialized using full backups.

Leave a Reply

Your email address will not be published. Required fields are marked *