Hello
I'm looking for a way of generating the next key value that works in MS and Sybase SQL Servers. Sybase identity columns are a bit dodgy, so...
If I have a separate table NextKey (NextKey int) with one row that I update as follows...
declare @NextKey int
update NextKey set NextKey = NextKey + 1, @NextKey = NextKey + 1
insert into myTable (PrimaryKeyCol, ....) values (@NextKey, ....)
are there any problems with concurrency ? As I see it the update will lock the row so different connections will always come up with a different @NextKey value....
So, I have some questions about best practice in SQL Server.
1.) I have PK like this (company TINYINT, store TINYINT, action TINYINT, invoice INT, sn SMALLINT). I know JOINS will work faster with surrogate key but I have only couple of JOINS on that table. I use members of PK in WHERE clause mainly, alone and combined for reporting purpose. Is it always better to have surrogate key because they don't have any meaning and context of data laying in current PK.
2.) In my PK from above I have two candidates for using Sequence object. Invoice start with 1 for every (company,store,action) combination. Sn start with 1 for every (company,store,action,invoice) combination. I would like to know can I implement Sequence object here knowing that Sequence don't support PARTITION BY in OVER clause. From what I red it cannot be done via Sequence but I have to ask.Here is data sample for this PK
hi! when i read some reference books about the SQL7.0, i often met 'surrogate key'. what's the surrogate key? what's its funtion? could you give me a good example? thanks very much!
The orininal design of my db (part of it...) is the following
A JOB has a Number and a Description. Each JOB can have one or two TASKS (min one, max two). Each TASK is identified by the JOB it belongs to and an Index (unique only for the same JOB). Each TASK has one an only one set of INFO1, one and only one set of INFO2, one and only one set of INFO3 etc.
(There is a reason to keep INFO1, 2 and 3 separate, because eachof them will be linked to different table. This might influence the answer to my real question.)
First of all, I wouldn't add any surrogate key for TASK, not to loose the logic behind; plus I'd put an ined on JonMum only, being Index equal to 1 or 2 only, so not selective.
The real question is about INFO1 (and 2, 3 etc.) table: should I leave JobNum and Index as PK (consider that the PK of INFo1 will be used as FK for another table), or should I use a surrogate key, like for eaxmple
C: INFO1 (Info1ID [PK], JobNum [FKb], Index [FKb], ...)
I don't really like this solution. Actually I'd prefer the following
C: INFO1 (Info1ID [PK], ...)
where Info1ID = JobNum + Index (+ = string concatenation).
Hey All, I'm trying to decide what's the 'best' to use. I've been designing and creating database for a while and have pretty much always used a surrogate key and not a normal one. I've finally had some free time to start studying more so in my spare time and read up and come accross a lot of guides, articles and stories that tout that normal keys should be used whenever possible as they're a better identifier and that surrogate keys should only be used when there is not a readily available normal key. Now perhaps I'd be open to accepting that but absolutely every database I come across tends to only use surrogate keys. For example I'm doing an authentication system from scratch and am looking at the User table. Now of course the user name has to be unique, should that be the primary key or should I have a seperate column with a guid or an incrementing int or the like as the primary key? I can certainly see that username could be used. I can also see how it may be easier when looking through the data tables to identify who/what a table is refering to with a surrogate key. However it still seems sort of sloppy, for lack of a better word, to me. Where now I could have somebody's username (or any other piece of data used for this purpose) spread accross a lot of other tables. And while writting this I just thought of the scenario that perhaps somebody needs their username changed, with this method now the ids need to be changed on all the related rows of all the other tables whereas with a surrogate key it wouldn't matter. Anyways I'm mostly looking for opinions on which way to go (not just with the user sample, but more in general).Thanks.
My previous post was not really clear, so I'll try again with a (hopefully) better (even if longer) example...
Consider the following...
A JOB describes the processment of a document. Each document can exist in two versions: English and French. A JOB can have 1 or 2 TASK, each describing the processement of either the English or French version. So we have the following:
that is there is an identifying 1:M (where maxium allowed for M is 2) relationship between JOB and TASK; TASK being identified by JobNum and Version (where the domain for Version is {E, F}).
Each TASK may require a TRANSLATION sub_task. Each TASK may require a TYPING sub_task. Each TASK may require a DISTRIBUTION sub_task.
For example, for a given doc, the English TASK requires TRANSLATION and DISTRIBUTION, while the French only DISTRIBUTION.
That is, there is a 1:1 not-required relationship between TASK and TRANSLATION, TYPING and DISTRIBUTION. So we have the following:
C: TRANSLATION (JobNum [PK] [FKb], Version [PK] [FKb], DueDate, ...) D: TYPING (JobNum [PK] [FKb], Version [PK] [FKb], DueDate, ...) E: DISTRIBUTION (JobNum [PK] [FKb], Version [PK] [FKb], Copies, ...)
As you can see I am using the PK of TASK as FK and PK for each of the three SUB_TASKs.
To complicate things, each SUB_TASK has one or more assignments. The assignments for each SUB_TASK records different information from the others. So we have...
C: TRANSLATION (JobNum [PK] [FKb], Version [PK] [FKb], DueDate, ...) D: TYPING (JobNum [PK] [FKb], Version [PK] [FKb], DueDate, ...) E: DISTRIBUTION (JobNum [PK] [FKb], Version [PK] [FKb], Copies, ...)
F: TRA_ASSIGN (JobNum [PK] [FKc], Version [PK] [FKc], Index [PK], Translator, ...) G: TYP_ASSIGN (JobNum [PK] [FKd], Version [PK] [FKd], Index [PK], Typyst, ...) H: REP_ASSIGN (JobNum [PK] [FKe], Version [PK] [FKe], Index [PK], Pages, ...)
that is there is an identifying 1:M relationship between each SUB_TASK and its ASSIGNMENTs, each ASSIGNMENT being identified by the SUB_TASK it belongs to and an Index.
I wish I could send a pic of the ER diagram...
Maybe there is another and better way to model this: if so, any suggestion?
Given this model, should I use for TRANSLATION, TYPING and DISTRIBUTION a surrogate key, instead of using the composite key, like for example:
C: TRANSLATION (TranslationID [PK], JobNum [FKb], Version [FKb], DueDate, ...) D: TYPING (TypingID [PK], JobNum [FKb], Version [FKb], DueDate, ...) E: DISTRIBUTION (DistributionID [PK], JobNum [FKb], Version [FKb], Copies, ...)
this will "improve" the ASSIGNMENTs tables:
F: TRA_ASSIGN (TranslationID [PK] [FKc], Index [PK], Translator, ...) G: TYP_ASSIGN (TypingID [PK] [FKd], Index [PK], Typyst, ...) H: REP_ASSIGN (DistributionID [PK] [FKe], Index [PK], Pages, ...)
I could even go further using a surrogate key even for TASK, which leads me to the following:
F: TRA_ASSIGN (TaskID [PK] [FKc], Index [PK], Translator, ...) G: TYP_ASSIGN (TaskID [PK] [FKd], Index [PK], Typyst, ...) H: REP_ASSIGN (TaskID [PK] [FKe], Index [PK], Pages, ...)
I don't really like this second solution, but I'm still not sure about the first solution, the one with the surrogate key only in the SUB_TASks tables.
I am performing a Select Into from a #table into a real table that has a surrogate key. If this is in a transaction (or not in one) am I guaranteed that the records inserted will be sequential surrogate key ids?
Select * into REALTABLE from MYPOUNDTABLE --40 rows
Can I assume that if the first one inserted is id 32 that the last one is 72?
{CREATE TABLEs and INSERTs follow...}Gents,I have a main table that is in ONE-MANY with many other tables. For example, ifthe main table is named A, there are these realtionships:A-->BA-->CA-->DA-->EWith one field in Common (Person). The tables B, C, D and E are History tables,with Start and End dates. Each person has a Program history (table B, ie), anExperience history (table C, ie), and so on...many differernt types ofhistories, and it may grow from here....table F, G, etc.The included CREATE TABLEs and INSERTs contain tables A, B and C.The problem: Each tblCase (table A) record has a date. When joining all of thehistory tables to tblCase on Person, obviously you get a cross-product of eachhistory unless you specify a WHERE clause that extracts one single record fromeach of the histories (duh...that's the point...to extract a single record fromeach history, because there can only be one value in effect at the time of theCase.)QUESTION: From a performance standpoint, would it behoove me to maintain thesurrogate ***HistoryID from each history table in tblCase, or, assuming theindexes are set up properly, would a WHERE condition for each history besufficient? For example, the following select works as expected:SELECT CasePerson, CaseDate, ProCode, ExpYearFROM tblExperienceHistory INNER JOIN (tblCase INNER JOIN tblProgramHistory ONtblCase.CasePerson = tblProgramHistory.ProPerson) ON tblCase.CasePerson =tblExperienceHistory.ExpPersonWHERE CaseDate BETWEEN ProStartDate and ProEndDateAND CaseDate BETWEEN ExpStartDate and ExpEndDateIt extracts the single record from each history for each person for each case.But I'm afraid of performace with such a scenario.Instead, I could store each ***HistoryID in the table tblCase, and then justjoin on that...no WHERE needed. But the trade-off is that I'd have to buildprocesses to maintain that. ("Hey, when you insert a record into tblCase, makesure to go get each HistoryID from the History tables!" or "If the user changesthe date ranges in one of histories, make sure to update tblCase to match thenew historyID!")Maybe a clustered index on each ***History table on Person/StartDate combinedwith the WHERE clause should perform as well as a real JOIN on surrogateintegers.It seems cheesey to have to resort to surrogate IDs...but the performanceincrease might be worth it. Also, if I go that route, whenever I add a newhistory table, I'd have to change the design of tblCase AND any SPs thatreference it. With the WHERE solution, I'd only have to change the SPs.Comments are welcome! (tblCase grows at 250,000 records per year; the historytables will increase about 1000 records per year)DCMFANCREATE TABLE [dbo].[tblCase] ([CaseID] [char] (5) CONSTRAINT [PK_tblCase] PRIMARY KEY CLUSTERED NOT NULL ,[CaseDate] [smalldatetime] NOT NULL ,[CasePerson] [char] (5) NOT NULL) ON [PRIMARY]GOCREATE TABLE [dbo].[tblExperienceHistory] ([ExperienceHistID] [int] IDENTITY (1, 1) NOT NULL ,[ExpPerson] [char] (5) NOT NULL ,[ExpStartDate] [smalldatetime] NOT NULL ,[ExpEndDate] [smalldatetime] NOT NULL ,[ExpYear] [int] NOT NULL) ON [PRIMARY]GOCREATE TABLE [dbo].[tblProgramHistory] ([ProgramHistID] [int] IDENTITY (1, 1) NOT NULL ,[ProPerson] [char] (5) NOT NULL ,[ProStartDate] [smalldatetime] NOT NULL ,[ProEndDate] [smalldatetime] NOT NULL ,[ProCode] [int] NOT NULL) ON [PRIMARY]GOINSERT INTO [tblCase]([CaseID], [CaseDate], [CasePerson])VALUES('12345', '3/1/03', '00000')INSERT INTO [tblCase]([CaseID], [CaseDate], [CasePerson])VALUES('A1G34', '4/23/03', '00001')INSERT INTO [tblExperienceHistory]([ExpPerson], [ExpStartDate], [ExpEndDate],[ExpYear])VALUES('00000', '1/1/03', '5/19/03', 1)INSERT INTO [tblExperienceHistory]([ExpPerson], [ExpStartDate], [ExpEndDate],[ExpYear])VALUES('00000', '5/20/03', '12/31/03', 2)INSERT INTO [tblExperienceHistory]([ExpPerson], [ExpStartDate], [ExpEndDate],[ExpYear])VALUES('00001', '4/20/03', '11/1/03', 0)INSERT INTO [tblProgramHistory]([ProPerson], [ProStartDate], [ProEndDate],[ProCode])VALUES( '00000', '2/1/03', '9/30/03', '55555')INSERT INTO [tblProgramHistory]([ProPerson], [ProStartDate], [ProEndDate],[ProCode])VALUES( '00000', '10/1/03', '5/1/04', '55555')INSERT INTO [tblProgramHistory]([ProPerson], [ProStartDate], [ProEndDate],[ProCode])VALUES( '00001', '1/1/03', '12/31/03', '55555')
I am performing a Select Into from a #table into a real table that has a surrogate key. If this is in a transaction (or not in one) am I guaranteed that the records inserted will be sequential surrogate key ids?
Select * into REALTABLE from MYPOUNDTABLE --40 rows
Can I assume that if the first one inserted is id 32 that the last one is 72?
Now, I like #1 because the implementation is simple -- the calling code simply passes an author name, and a country id and an INSERT INTO statement is called with those parameters
INSERT INTO authors( @authorName, @countryId )
I like #1, because it hides the surrogate "id" key from the application calling code. But on the downside, it has more overhead work, because you have to first a) verify a country with that name exists, and b) select that id into a variable.
DECLARE id INT; IF EXISTS (select * from countries where country_id = @countryId ) THEN SELECT country_id INTO id FROM countries WHERE country_name = @countryName; END IF;
(Sorry I may have the SQL syntax wrong up there, but I was just trying to demonstrate the extra overhead involved).
I am in the process of building a fact table in a staging area. The data in the host system has numerous composite keys, so I have replaced all the composite keys in the dimensions with surrogate keys (integer) which are generated using an identity at load time. When I load the staging (fact) table, I have set the default value of all the foreign keys to 0. What I must do now is update all the foreign key values with the surrogate key values from the dimensions. I'm using an update command and the original gid values from the source system in the where clause...i.e. UPDATE X SET x.key_1 = y.key_1 FROM TableA X WITH (NOLOCK) INNER JOIN TableB Y WITH (NOLOCK) ON x.org_id = y.org_id AND x.bus_id = y.bus_id AND x.prov_gid = y.prov_gid AND x.log_gid = y.loc_gid;
This seems to work fine for most tables. However, I am now trying to update a table that has over 10 million rows and approximately 30 foreign keys. The script runs for hours. I ususally stop it after about 8 hours when it still hasn't completed. Since the keys are dynamic and they could possibly change during each load process, I can't add them during the load process.
Is there a better way to update these keys. I need to regenerate the fact tables every night and taking this much time to reload a fact table is just not practicle. I've indexed the alternate keys on all the dimensions and have also indexed the gids on the target fact table. Am I doing something wrong? Have I over indexed the target table? Please help! Thanks Jerry
I have a dimension called 'Caller Type' with the following attributes:
CallerTypeKey ---- surrogate key
CallerTypeID
CallerTypeDesc
CreatedByKey ---- foreign surrogate key from User Dimension
I used Script Task to get the last used key and increment it so i can use it for new records in my dimension. however, my dimension is linked to a User Dimension and I need the surrogate key of that once I insert the new record to CallerType Dimension.
This is the code iam using to get the incremental surrogate keys:
Imports System Imports System.Data Imports System.Math Imports Microsoft.SqlServer.Dts.Pipeline.Wrapper Imports Microsoft.SqlServer.Dts.Runtime.Wrapper
Public Class ScriptMain Inherits UserComponent 'Declare a variable scoped to class ScriptMain Dim counter As Integer
Public Sub New() 'This method gets called only once per execution 'Initialise the variable counter = 1093 End Sub
'This method gets called for each row in the InputBuffer Public Overrides Sub Input_ProcessInputRow(ByVal Row As InputBuffer) 'Increment the variable counter += 1
'Output the value of the variable Row.instance = counter End Sub
End Class
--'Instance' is my surrogate feild name
but iam getting an error saying that InputBuffer is not defined ..Any idea?
If I want to add two more incrementive fileds ,where i have to add it?
Sorry if it sounds silly ,iam very new to this scripting.
I have a database surrogate key that increments so rapidly (+5000 every 30 mins). I need my SSIS package to reset this database surrogate key to avoid reaching an upper limit value for that field.
I have several stage to star (i.e. moving data from a staging table through the key lookups into a fact table) ETL transformations in a single SSIS package. Each fact table has a different set of measures but the identical foreign key set, e.g. ConsultantKey, SubsidiaryKey, ContestKey, ContestParamKey and MonthKey.
Currently I have to replicate the key lookup (Surrogate Key Pipeline, or SKP) for each data flow. If I could cache each dimension one time in the package and reuse it for each stage to fact it would be much more efficient.
Is there a way for me to reuse a common data flow?
I am trying to write a ssis surrogate key data transform, my problem is I can't find an example how to add a column to the incoming columns and add some data to it. If anyone has a sample, can you please post it. I found a script option that works but I would like an actual transform.
I want to change the work table name to work_version2 and later drop the work table. First, I created the table (work_version2) along with the data structure seen below and later inserted data from the work table. As I tried to make workID a surrogate key in work_version2 using SSMS, I got the below error message when I try to save the changes. Is there a way to do this?
Saving changes is not permitted. The changes you have make requires the following tables to be dropped and re-created. You have either make changes to a table that cant't be recreated or enabled the option that prevent saving changes that requires the table to be recreated. Work_version2.
CREATE TABLE WORK( WorkID Int NOT NULL IDENTITY (500,1), Title Char(35) NOT NULL, Copy Char(12) NOT
I have tables like the one below for my Stage and dimension tables:
Stage Table
accountid
name
address
Dimension Table
accountkey ---- surrogate key (DW key)
accountid ---- business key (transaction's primary key)
name
address
I used slowly changing dimension to detect the changes for the records inside my Dimension table. But I had a problem when a new record exists in the stage table. The accountkey is set as the primary key and it gets its value from a different table which stores the last account key that was created. I cannot load all the changes unless i have a business key. Is there a way that i can get the "last key" from a different table in the data flow area and then supply it together with the other fields in the new output branch of the slowly changing dimension?
I want to create an import table for daily rows with an integer column like 20150430 for the date, called DayKey. This table would do one date per day. It would then be imported into a STAGE table which would have the same columns and would have all of the import rows for every day.My question would be this: I want to be able to have an integer Primary Key unless there is a better idea. I could make the STAGE table use an auto-incremented value for the key. Then, when I load the import table which is truncated every day, I could take the NEXT value of the key from the STAGE table and increment by 1.
Let's say the last value in STAGE is 1000, then the next value that would be in IMPORT would be 1001 and incrementing up. Then these would be added to the STAGE table with the associated keys. There is no chance of anyone or anything else adding to the STAGE table any other way.
Hi, I use lookups to map surrogate of level 1 dimensions to my fact tables in SSIS. But how to handle a level 2 dimension with a ValidFrom and a ValidUntil date field? I do not use an IsCurrent column, because this could problem with late arriving facts.
- In dts I used an SQL statement like this:
update SA SET SA.DimProdRef = Dim.RecordID FROM SAWarenEingang SA, DimProd Dim where SA.ProduktNumber = Dim.ProduktNumber and SA.ArtikelkontoBewegungsdatum between Dim.ValidFrom and Dim.ValidUntil
Now in SSIS I want to handle the whole thing in the data flow without using a staging table: - Using Lookups: I would have to pass the date column for each inside the fact table into the lookup. That does not work. - Using Execute SQL in the data flow: would be very slow, because the statement will be executed for any line in the dataflow
My question here is as the dimension has SCD type 2 on it and every time when there is a change the persn_key gets a new key value but the fact table still points to oldest key.how to update the surrogate key on fact table to the current key value? As per the requirement fact surrogate key must be pointing to current active record on the dimension.
While I have learned a lot from this thread I am still basically confused about the issues involved.
.I wanted to INSERT a record in a parent table, get the Identity back and use it in a child table. Seems simple.
To my knowledge, mine would be the only process running that would update these tables. I was told that there is no guarantee, because the OLEDB provider could write the second destination row before the first, that the proper parent-child relationship would be generated as expected. It was recommended that I create my own variable in memory to hold the Identity value and use that in my SSIS package.
1. A simple example SSIS .dts example illustrating the approach of using a variable for identity would be helpful.
2. Suppose I actually had two processes updating these tables, running at the same time. Then it seems the "variable" method will also have its problems. Is there a final solution other than locking the tables involved prior to updating them or doing something crazy like using a GUID for the primary key!
3. We have done the type of parent-child inserts I originally described from t-sql for years without any apparent problems. (Maybe we were just lucky.) Is the entire issue simply a t-sql one or does SSIS add a layer of complexity beyond t-sql that needs to be addressed?
I want to insert a new record into a table with an Identity field and return the new Identify field value back to the data stream (for later insertion as a foreign key in another table).
What is the most direct way to do this in SSIS?
TIA,
barkingdog
P.S. Or should I pass the identity value back in a variable and not make it part of the data stream?
I have table of three column first column is an ID column. However at creation of the table i have not set this column to auto increment. Then i have copied 50 rows in another table to this table then set the ID column values to zero.
Now I have changed the ID column to auto increment seed=1 increment=1 but the problem is i couldn't figure out how to update this ID column with zero value set to each row with this auto increment values so the ID column would have values from 1-50. Is there a away to do this?
Ok,I just need to know how to get the last record inserted by the highestIDENTITY number. Even if the computer was rebooted and it was twoweeks ago. (Does not have to do with the session).Any help is appreciated.Thanks,Trint
Hi, I am having problem in bulk update of a sql server table haning identity column from a datatable( has no identity column) using sqlbulkcopy. I tried several approaches, but it does not show any error nor is the table getting updated. But the identity value seems to getting increased every time. thanks. varun
I'm working with a third-party database (SQL Server 2005) and the problem here is the following:
- There are a bunch of ETL processes that needs to insert rows on a table (let's call this table T) and at the same time, an ERP (owner of T) is up and running (reading, updating and inserting on T).
- The PK of T is an Integer.
Today all ETL processes uses (select max(ID) + 1 from T) to insert new rows, so just picture the scenario. It is a mess! Everyday they get duplicate key error when 2 or more concurrent processes are trying to insert a row (with the max) at the same time.
Considering that I can't change the PK, what is the best approach to solve this problem?
To sum up:
* I need to have processes in parallel inserting on T
I am developing a ASP.NET application using SQL Server as my database. I want to create a number that is unique.
The number will have 8 digits.
first 2 digits correspond to year. ex:04 or 03
the next 6 digits start with 000001 and it should get added for each new entry of data.
i am not able to generate number in the way i said. I am relatively new to SQL. so any suggestions as how to go about solving the problem???. Are there any samples/codes available for this.
I have generated an SQL script such that when it is run, it will automatically create tables if not existing or drop the existing tables. The problem I have is that, is there any way I can also make the script create not only the tables but also transfer the data from the old database to the new database?
I would really appreciate if any of you could point me to any existing programs, scripts etc that can do that. Thanks in advance
