Comments on: SQL SERVER – Negative Identity Seed Value and Negative Increment Interval

By: SQL SERVER – RESEED Identity Column in Database Table – Rest Table Identity Value – SQL in Sixty Seconds #051 | SQL Server Journey with SQL Authority

Wed, 08 May 2013 01:31:31 +0000

[...] Negative Identity Seed Value and Negative Increment Interval [...]

By: Akash

Akash — Tue, 27 Mar 2012 08:42:00 +0000

I have an application in which a task can be assigned to either a specific user or a user group. The user table has a primary key with a positive increment starting from 1, the user group table has a primary key with a negative increment starting from -1. If the task references a negative ID, I know it has been assigned to a user group.

You may argue that this is crummy design (and you’d probably be right), but the concept of user groups was added long after the application had been delivered and this was the quickest way to implement it…

By: Amir Oveissian

Amir Oveissian — Thu, 26 Jan 2012 14:38:26 +0000

This was very useful, Saved me half a day checking 2 tables for conflict in IDs to show them in 1 interface grid!!!!

By: SQL SERVER – Understanding Identity Beyond its Every Increasing Nature – Quiz – Puzzle – 3 of 31 « SQL Server Journey with SQL Authority

Wed, 04 Jan 2012 01:32:55 +0000

[...] Gap in Identity Column Discussion – Effect of Missing Identity on System – Real World Scenario Negative Identity Seed Value and Negative Increment Interval Enable Identity Insert – Import Expert Wizard Denali – SEQUENCE is not [...]

By: Kevin R

Kevin R — Mon, 01 Aug 2011 00:27:53 +0000

Metric SI prefixes table:

ID NAME SYMBOL
-9 nano n
-6 micro µ
-3 milli m
3 kilo k
6 mega M
…

By: Pramod

Pramod — Thu, 02 Dec 2010 06:20:58 +0000

Hello Pinal Sir,

I have done some arithmetic operation on table fields using select query and insert that result into anther table,

But Problem is If i subtract two fields then some result getting negative value just i have check condition if result is negative then inset zero value in that row and column

Please suggest me
How can Possible it

By: Priyadarshi Alok

Priyadarshi Alok — Thu, 09 Sep 2010 09:09:51 +0000

nice example

By: Mitesh

Mitesh — Wed, 25 Aug 2010 21:00:24 +0000

It is often used in peer-to-peer replication, where one server increments positively and other negatively, thus avoiding the conflicts when the sync up occurs.

By: Marko Parkkola

Marko Parkkola — Fri, 20 Aug 2010 11:42:05 +0000

For same reason you can’t use order by clause why creating a table ;) You order it when you use it.

What you can do is schemabind the view and add clustered index to the field you want to order it by. Just take notice that not all views can be schemabound or indexed and indexed view lowers the performance of insert/update/delete operations.

By: Calvin

Calvin — Thu, 19 Aug 2010 20:55:44 +0000

We have an application where some users can fill out a questionnaire that gets stored to a database inside the firewall and some users can fill out a questionnaire using a public facing web app that get stored to a database outside the firewall. The questionnaire records from outside the firewall have negative identity values and get replicated to the database inside the firewall whose records have positive identities.

By: Ramdas

Ramdas — Thu, 19 Aug 2010 17:31:09 +0000

Interesting take on negative identity values, always used the default positive values… Thanks for the link BEL8490

By: Nakul Vachhrajani

Nakul Vachhrajani — Thu, 19 Aug 2010 17:02:54 +0000

Hello!

I did know that a negative identity value can be used (tried that myself a long time ago) – however, I have been thinking about it but have failed to come up with a practical use of the same for data storage.
However, my belief is that if Microsoft has provided some feature, they definitely use it, and it’s only a matter of time and research before we find where.
My guess is that such a feature would be used when we are trying to decrement age (eg. lazy writer or applications that count towards 0 upon occurance of an event). But, that’s only a guess.

Have a great day, everyone!

By: mjfii

mjfii — Thu, 19 Aug 2010 14:24:10 +0000

I always seed my identity columns with the smallest value possible to get the full use out of the data type, i.e. negative numbers.

create table #x
(
x smallint identity(-32768,1)
)

create table #x
(
x int identity(-2147483648,1)
)

The smallint identity gives you the use 65535 identities, which under many business rules is plenty of tuples to play with and saves space (just a little)…

mjfii

By: Luke L

Luke L — Thu, 19 Aug 2010 14:16:10 +0000

Best Business reason I’ve seen is that you know you need more than 2 billion identity values but are unlikely to ever need 4. Think a read only historical database or something along those lines. Starting with a negative seed and working your way up allows you to use the more compact int datatype instead of bigint.

By: krazykooter

krazykooter — Thu, 19 Aug 2010 13:42:46 +0000

I’ve seen this used where a value was needed at the beginning of an existing code table. Since the existing code table had the identity value start a 1, then 0 and -1 was used for the new values that were to be at the beginning.

By: fly

fly — Thu, 19 Aug 2010 13:24:49 +0000

needed it once when i had to merge two identical tables from different DB’s.
Set identity insert on and adding “-” sign prevents from duplicate identities.

By: BEL8490

BEL8490 — Thu, 19 Aug 2010 07:30:13 +0000

Negative identity values are useful in programming: if you have build part of your database in memory as a dataset and you want to submit the changes back to the server, you cannot be sure of the identity values it will assign (especially in a shared environment). Therefore it is common to use negative identities and increments for the memory dataset and positive ones for the server database. The server does understands that mechanism and will assign correct values (It will take some tricks to refresh these values into the memeory dataset).
For a more elaborate artikel see: Managing an @@Identity crisis at http://msdn.microsoft.com/en-us/library/ms971502.aspx

By: Sonia

Sonia — Thu, 19 Aug 2010 06:42:29 +0000

Awsome example… I never thought in this direction…

By: Anil Vanjre

Anil Vanjre — Thu, 19 Aug 2010 05:48:34 +0000

Hi Pinal Dave,

Could you please tell me why we can’t use

Order By Cluase , While Creating a view….?

By: Jeff Stryker

Jeff Stryker — Thu, 19 Aug 2010 05:04:49 +0000

Great post. I wouldn’t have guessed this was possible. You are a SQL authority.

By: kalyanasundaram.k

kalyanasundaram.k — Thu, 19 Aug 2010 03:58:27 +0000

Hi,

I check thie Negative Identity in my Server, Its Nice.

I dont think like that, whether its possible or not.

But i try to reseed the table, it works fine.

create a table :

CREATE TABLE dbo.IdentityNegative
(
ID INT NOT NULL IDENTITY (1, 1),
Firstcol VARCHAR(100) NULL
)

Insert few records, then reseed the identity

dbcc checkident (‘dbo.IdentityNegative’,'reseed’,0)

Same like as

CREATE TABLE dbo.IdentityNegative
(
ID INT NOT NULL IDENTITY (-1, -1),
Firstcol VARCHAR(100) NULL
)

Insert few records, then reseed the identity

dbcc checkident (‘dbo.IdentityNegative’,'reseed’,0)

Here, i will use the same RESEED query for both negative and positive identity tables. Its identify the table have positive or Negative Identity?

How this action is performed in SQL Server?