Manage and Monitor Identity Ranges in SQL Server Transactional Replication

Problem

When using transactional replication to replicate data in a one way topology from a publisher to a read-only subscriber(s) there is no need to manage identity ranges. However, when using  transactional replication to replicate data in a two way replication topology - between two or more servers there is a need to manage identity ranges in order to prevent a situation where an INSERT commands fails on a PRIMARY KEY violation error  due to the replicated row being inserted having a value for the identity column which already exists at the destination database.

Solution

There are two ways to address this situation:

  1. Assign a range of identity values per each server.
  2. Work with parallel identity values.

The first method requires some maintenance while the second method does not and so the scripts provided with this article are very useful for anyone using the first method. I will explore this in more detail later in the article.

In the first solution set server1 to work in the range of 1 to 1,000,000,000 and server2 to work in the range of 1,000,000,001 to 2,000,000,000.  The ranges are set and defined using the DBCC CHECKIDENT command and when the ranges in this example are well maintained you meet the goal of preventing the INSERT commands to fall due to a PRIMARY KEY violation. The first insert at server1 will get the identity value of 1, the second insert will get the value of 2 and so on while on server2 the first insert will get the identity value of 1000000001, the second insert 1000000002 and so on thus avoiding a conflict.

Be aware that when a row is inserted the identity value (seed) is generated as part of the insert command at each server and the inserted row is replicated. The replicated row includes the identity column’s value so the data remains consistent across all servers but you will be able to tell on what server the original insert took place due the range that  the identity value belongs to.

In the second solution you do not manage ranges but enforce a situation in which identity values can never get overlapped by setting the first identity value (seed) and the increment property one time only during the CREATE TABLE command of each table.

So a table on server1 looks like this:

CREATE TABLE T1 
(
 c1 int NOT NULL IDENTITY(1, 5) PRIMARY KEY CLUSTERED
,c2 int NOT NULL
);

And a table on server2 looks like this:

CREATE TABLE T1
(
 c1 int NOT NULL IDENTITY(2, 5) PRIMARY KEY CLUSTERED
,c2 int NOT NULL
);

When these two tables are inserted the results of the identity values look like this:

Server1:  1, 6, 11, 16, 21, 26…

Server2:  2, 7, 12, 17, 22, 27…

This assures no identity values conflicts while leaving a room for 3 additional servers to participate in this same environment. You can go up to 9 servers using this method by setting an increment value of 9 instead of 5 as I used in this example.

Continues…

Pages: 1 2




Related Articles :

No comments yet... Be the first to leave a reply!

Software Reviews | Book Reviews | FAQs | Tips | Articles | Performance Tuning | Audit | BI | Clustering | Developer | Reporting | DBA | ASP.NET Ado | Views tips | | Developer FAQs | Replication Tips | OS Tips | Misc Tips | Index Tuning Tips | Hints Tips | High Availability Tips | Hardware Tips | ETL Tips | Components Tips | Configuration Tips | App Dev Tips | OLAP Tips | Admin Tips | Software Reviews | Error | Clustering FAQs | Performance Tuning FAQs | DBA FAQs |