|
Transaction log entries are made whenever you make a change to your tables: perform an UPDATE, DELETE or INSERT operation. The log entries are required to allow you to roll back your transaction. They are required to allow an administrator to roll forward operations from the last full backup, if, for example, a disk failure occurs. They also permit the database to be rolled back to a specific point in time, if, say, a coding error causes data to be deleted. Finally they're used during database startup recovery, if the database was not cleanly shut down (e.g. system power loss, SQL Server crash).
The transaction log contains the before and after image of each row affected: in the case of an INSERT there is obviously no 'before' image and for a DELETE no 'after' image. To constrain the amount of transaction log required and therefore the amount of transaction log I/O, you need to reduce the number of rows you affect, if possible, and ensure that you update each row as few times as possible.
If you've already done this, and you're sure that transaction log I/O is your bottleneck, you need to take steps to ensure that the physical transaction log I/O is as fast as possible. Firstly, ensure it's on a separate physical disk or array of disks to anything else - the transaction log requires almost entirely sequential write with very occasional reads required when a transaction is rolled back, whereas the database data files are fully random reads and writes. Mixing random and sequential I/O sends the disk head out of position to perform the next sequential I/O, which increases the time taken to write.
Because the transaction log is so vital to the integrity of the database, you should ensure that any RAID array you use has some form of integrity checking to allow the array to be reconstructed on a disk failure: one of RAID 1, RAID 5 or RAID 10. Of these, RAID 5 is generally considered weakest - RAID 5's gains are in random read performance because random I/Os are spread across disks, while all writes must read from all disks (to compute the checksum) and write to two disks (the actual data, and the computed checksum). You can get better write performance from RAID 1 (mirroring, all writes go to two disks), and RAID 5's performance suffers badly when rebuilding an array. Likewise, although RAID 10 allows improved random-access read and write performance, it does nothing for sequential reads or writes.
You may be able to improve performance by using a battery-backed write-back caching disk controller. If the controller guarantees to write back operations in the same order that they were received by the controller, and you are sure that mains power will be restored to the server before the controller's battery runs out, it's safe. If the controller, or you, can't meet these guarantees, you could suffer database corruption. Search Microsoft's knowledge base for "cache controller sql" for more details.
Stability. What an interesting concept. -- Chris Maunder
|
|
|
|
|
i am writing a query where i need to find two dates which are exactly one month and two months less than a given date.
like if date is '2006/05/01'
than result dates should be '2006/04/01' and '2006/03/01' respectively.
pls help
|
|
|
|
|
Use DATEADD[^]
For example:
SELECT DATEADD(MONTH, -1, TheDate), DATEADD(MONTH, -2, TheDate)
FROM SomeTable
Does this help?
"On two occasions, I have been asked [by members of Parliament], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able to rightly apprehend the kind of confusion of ideas that could provoke such a question."
--Charles Babbage (1791-1871)
My: Website | Blog
|
|
|
|
|
|
Hi,
I create my installation project using VS.NET 2005 , then I add the merge modules to the project ! but it doesn't work !
- I also tried to customize the MSDE Internal properties but it didn't work !
any suggestions !!!
hatem++;
|
|
|
|
|
From what I recall, using the merge modules is now not recommended. Microsoft continue to ship the merge modules for applications already using them. See this article[^], specifically the comment at the end:
"PSS Recommendations Regarding MSDE Merge Modules
"Due to rapidly changing threats to the security of our IT systems, we strongly suggest that customers use the MSIs provided by Microsoft and not use merge modules. Using the MSIs provided by Microsoft allow the customer greater flexibility in responding to these threats and making sure that their MSDE embedded applications are secure."
You should consider launching the MSDE setup from your own setup launcher program. You may be able to do this with a custom action. See also Microsoft's MSDE 2000 Deployment Toolkit[^].
IIRC the recommended practice for using SQL Server 2005 Express Edition is to allow setup to create a single SQLEXPRESS instance if one is not already present. All applications should then share that instance, specifying the database file to use in the connection string. Microsoft have belatedly realised that this isn't terribly convenient for developers and are now discussing porting SQL Server 2005 Mobile Edition to Windows XP and Windows Server 2003, from Windows CE - it will be called 'SQL Server Everywhere Edition'. This will be implemented as a set of DLLs and will run in your application's process, rather than as a separate service. I suspect this is the long-term replacement for the Jet database engine used by Access.
Stability. What an interesting concept. -- Chris Maunder
|
|
|
|
|
I would like to know how to rename a certain table through sql script.
I tried to do a stored procedure with the following commands where NewName represents the new table name and PreviousName represents the previous name :
UPDATE [TDM-DataSets].INFORMATION_SCHEMA.TABLES
SET TABLE_NAME = @NewName
WHERE TABLE_NAME = @PreviousName;
The error message i got is :
Ad hoc updates to system catalogs are not allowed.
Does anyone have an idea ?
Thanks,
Clint
|
|
|
|
|
"Rename Table @PreviousName TO @NewName"
We need to graduate from the ridiculous notion that greed is some kind of elixir for capitalism - it's the downfall of capitalism. Self-interest, maybe, but self-interest run amok does not serve anyone. The core value of conscious capitalism is enlightened self-interest.
Patricia Aburdene
Bulls make money, bears make money, pigs get slaughtered.
Jim Cramer
|
|
|
|
|
I put the line in the stored procedure and it doesn't work.
I am working with SQL Server.
it doesn't recognize the command rename.
Thanks,
Clint
-- modified at 11:00 Sunday 30th April, 2006
|
|
|
|
|
Thanks,
I found out :
EXEC dbo.sp_rename 'OldTableName', 'NewTableName'
|
|
|
|
|
Hi,
I have one ObjectDataSource which has a Select query that requires Parameter @CompanyId. CompanyId is Guid and I want to compare it with a bunch of GUID. Can i pass those Guid programatically???
my Query is like..
select a.companyId, b.startdate, b.....
from a,b
where a.companyId IN (@CompanyID)
|
|
|
|
|
hi there
yes you can use this query but remember to pass @CompanyID as a comma separated string of ids.
|
|
|
|
|
Hi,
Thanks for the answer. I found one alternate way to do it as someone suggested me from Asp.net forum.
But, Your reply made me think about the simple mistake i was making. I had provided Comma between values but I did not provide quotes between value. Hence, each guid was compared to entire string i suppose. I'll try to put 'guid','guid' format as well.
Thanks
|
|
|
|
|
in SqlExpress!!!
i created a db in VS.NET 2005, and i want to change the collation of the data base to another language, but it seems it is disabled and set to " <database defalut=""> ".
so what should i do?
Militiaware
Faris Madi
|
|
|
|
|
|
i wanna create database on VS.Net 2006 SQLExpress.
an error ocuree telling me that the default setting of sqlserver can't connect remotly.
and i don't wanna connect remotly.
all services running...
and when i create new driver in Data Sources (ODBC) he is keep telling me that the SQL Server not exist.
so what i can do!?
it's urgent
Militiaware
Faris Madi
|
|
|
|
|
Try using SQLExpress and VS.NET2005. There may be compatibility issues with version 2006...
Pictures of the menu available at the drive-thru
|
|
|
|
|
Why are you using ODBC when there is a built in .NET Data Provider for SQL Server?
"On two occasions, I have been asked [by members of Parliament], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able to rightly apprehend the kind of confusion of ideas that could provoke such a question."
--Charles Babbage (1791-1871)
My: Website | Blog
|
|
|
|
|
the SQL SERVER driver don't exist from the begining!!1
Militiaware
Faris Madi
|
|
|
|
|
militiaware wrote: the SQL SERVER driver don't exist from the begining!!1
What?! That doesn't make sense! What SQL Server driver are you talking about? If you are referring to ODBC use the SQL data provider built into .NET (you can find it in the System.Data.SqlClient namespace)
"On two occasions, I have been asked [by members of Parliament], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able to rightly apprehend the kind of confusion of ideas that could provoke such a question."
--Charles Babbage (1791-1871)
My: Website | Blog
|
|
|
|
|
Maybe it doesn't come with the VS2006 version he says he is using. 
Pictures of the menu available at the drive-thru
|
|
|
|
|
|
 I am using sql 2000. I have a scenario is 1 user can buy many or 0 'package'.
and 1 'package' can have 0 or many users. If the user didnt buy the 'package' then he/she will not allow to use the package.
How should I create my database. How is my database look like.
|
|
|
|
|
In other words a many-to-many join
Here are the key columns in use and their tables
User: UserID
Package: PackageID
UserPackage: UserID, PackageID
The UserPackage table is the intermediate table in the many-to-many join. It has a compound key consisting of both sets of primary keys. When you need to join a user to a package you put a row in the UserPackage table.
Does this help?
"On two occasions, I have been asked [by members of Parliament], 'Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?' I am not able to rightly apprehend the kind of confusion of ideas that could provoke such a question."
--Charles Babbage (1791-1871)
My: Website | Blog
|
|
|
|
|
I am actually creating my project in .net.
ok now i have my sql command as like this
SELECT * FROM t_userPackage INNER JOIN t_package,t_user ON t_userPackage.packageID = t_package.packageID AND userID = '" & User.Identity.Name & AND t_package.packageID=XXXXXXXXX
for XXXXXXXX what should I put? my program is when you click on the list, it will then run the sql command to check whethere the specific user have register/pay for the specific package.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
Of course! I completely forgot about that! 
