Efficient way to get the lowest unused value of a nvarchar field being used as a primary key

  • Hi all,

    Trying to get the lowest unused value of CrewID from the below table in a efficient manner but stumped and was wondering if any one could think of a good solution. I cant change the table structure as its from a 3rd party and it would break our support agreement.

    CREATE TABLE [dbo].[Crews](

    [CrewID] [nvarchar](10) NOT NULL,

    [CrewName] [nvarchar](200) NOT NULL,

    [CompanyID] [nvarchar](10) NOT NULL,

    [StartTime] [nvarchar](50) NULL,

    [BreakTime] [real] NULL,

    [FinishTime] [nvarchar](20) NULL,

    [Notes] [ntext] NULL,

    [IsPartial] [smallint] NOT NULL,

    [LastModified] [datetime] NOT NULL,

    [LastModifiedBy] [nvarchar](50) NULL,

    [GUID] [uniqueidentifier] NOT NULL,

    CONSTRAINT [PK_Crews] PRIMARY KEY CLUSTERED

    (

    [CrewID] ASC

    )WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]

    ) ON [PRIMARY] TEXTIMAGE_ON [PRIMARY]

    GO

    ALTER TABLE [dbo].[Crews] ADD CONSTRAINT [DF_Crews_BreakTime] DEFAULT (0) FOR [BreakTime]

    GO

    ALTER TABLE [dbo].[Crews] ADD CONSTRAINT [DF_Crews_IsPartial] DEFAULT (0) FOR [IsPartial]

    GO

    ALTER TABLE [dbo].[Crews] ADD CONSTRAINT [DF_Crews_LastModified] DEFAULT (getdate()) FOR [LastModified]

    GO

    ALTER TABLE [dbo].[Crews] ADD CONSTRAINT [DF_Crews_GUID] DEFAULT (newid()) FOR [GUID]

    GO

  • How do you know when a CrewID value has been 'used'?

    Regards

    Lempster

  • Hi Lempster,

    People can put whatever they want in there unfortunately. There no really way without going through all the data that i can think of. I personally think its bad design but have to work with it unfortunately.

    Regards,

    Michael

  • micalo (3/21/2014)


    Hi Lempster,

    People can put whatever they want in there unfortunately. There no really way without going through all the data that i can think of. I personally think its bad design but have to work with it unfortunately.

    Regards,

    Michael

    The lowest unused value of an nvarchar column...what characters would you like to take into account?

    “Write the query the simplest way. If through testing it becomes clear that the performance is inadequate, consider alternative query forms.” - Gail Shaw

    For fast, accurate and documented assistance in answering your questions, please read this article.
    Understanding and using APPLY, (I) and (II) Paul White
    Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff Moden

  • can you show Query, how did you try to pick the lowest value

    Every rule in a world of bits and bytes, can be bend or eventually be broken
    MyBlog About Common dialog control
    A Visualizer for viewing SqlCommand object script [/url]

  • My best quick guess is that a StartTime of NULL indicates an unused entry. Even if not, the code below may give you some ideas for the correct column(s) on which unused status is based.

    --first create an index to speed lookup

    CREATE UNIQUE NONCLUSTERED INDEX IX_Crews_StartTime

    ON dbo.Crews ( StartTime, CrewID )

    --WHERE valid only on SQL 2008+

    WHERE StartTime IS NULL

    --if on Enterprise Edition, uncomment "ONLINE = ON,"

    WITH ( FILLFACTOR = 99, /*ONLINE = ON,*/ SORT_IN_TEMPDB = ON )

    ON [PRIMARY]

    --then do the actual lookup itself

    SELECT /*@CrewID = */ MIN(CrewID)

    FROM dbo.Crews

    WHERE StartTime IS NULL

    SQL DBA,SQL Server MVP(07, 08, 09) A socialist is someone who will give you the shirt off *someone else's* back.

  • Hi Scott,

    Thanks for trying but didn't work unfortunately. I had another crack at it but in the end decided the that its really in the best interest of the company that developed the app to have a identity as the PK. Will just have to go through the experience of explaining why but it think its worth it for them and for me. There'll be some benefits for them in performance as well where this is used in joins if the do that and if they have that in one table it might be worth taking a look at the whole db and offering some suggestions for everyone's sake. Wanted to give it a go as I like a challenge.

    Thanks,

    Michael

  • FYI,

    before deciding its not worth it this is the sql I was playing with. It works but i really didn't want to show it to anyone a I think its a bad solution :ermm:

    DECLARE @ID varchar(10)

    WHILE @1<=126

    BEGIN

    WHILE @2<=126

    BEGIN

    WHILE @3<=126

    BEGIN

    WHILE @4<=126

    BEGIN

    WHILE @5<=126

    BEGIN

    SELECT @ID = 'CREW' + ltrim(char(@1)) + ltrim(CHAR(@2)) + ltrim(char(@3)) + ltrim(CHAR(@4)) + ltrim(char(@5))

    IF (SELECT COUNT(CrewID) FROM Crews WHERE CrewID =@ID)=0

    BEGIN

    GOTO ADDME;

    END

    SELECT @5=@5+1

    END

    SELECT @5=32

    SELECT @4=@4+1

    END

    SELECT @4=32

    SELECT @3=@3+1

    END

    SELECT @3=32

    SELECT @2=@2+1

    END

    SELECT @2=32

    SELECT @1=@1+1

    END

    ADDME:

    SELECT @ID

  • Hi there,

    Thanks, I think you try to pick the missing crewid well need some more data, minimum crewid maximum crewid and some sample crewid to know how the crewid is formed, because from your code the logic is missing length, of the column is 10, from your code it may reach 11, please post some data to identify the pattern of the crewid

    Every rule in a world of bits and bytes, can be bend or eventually be broken
    MyBlog About Common dialog control
    A Visualizer for viewing SqlCommand object script [/url]

  • micalo (3/23/2014)


    Hi Scott,

    Thanks for trying but didn't work unfortunately. I had another crack at it but in the end decided the that its really in the best interest of the company that developed the app to have a nvarchar as the PK. Will just have to go through the experience of explaining why but it think its worth it for them and for me. There'll be some benefits for them in performance as well where this is used in joins if the do that and if they have that in one table it might be worth taking a look at the whole db and offering some suggestions for everyone's sake. Wanted to give it a go as I like a challenge.

    Thanks,

    Michael

    "its really in the best interest of the company that developed the app to have a nvarchar as the PK"

    Can you explain why? If you're going to use a surrogate key, surely an identity column is better than an arbitrary nvarchar value containing tab, space etc characters and which you have to generate for every row?

    "There'll be some benefits for them in performance as well"

    Can you elaborate? Benefits compared to what?

    Why include "Crew" as a suffix in the column values? Call the column 'CrewID' and gain 4 character spaces - you could probably get away with a varchar column instead of nvarchar. Then you won't have to jump through hoops figuring out a slow and dodgy algorithm to find unused PK's.

    If you like a challenge, and offering a challenge to anyone else who's going to be working on this database, then you've chosen well.

    “Write the query the simplest way. If through testing it becomes clear that the performance is inadequate, consider alternative query forms.” - Gail Shaw

    For fast, accurate and documented assistance in answering your questions, please read this article.
    Understanding and using APPLY, (I) and (II) Paul White
    Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff Moden

  • Hi Chris,

    lol it was a typo, it should have said identity. sorry for the confusion

Viewing 11 posts - 1 through 10 (of 10 total)

You must be logged in to reply to this topic. Login to reply