Category Archives: Data

Why You Can’t Directly Bulk Load Values
or See Them in Traces

,

Bulk loading seems shrouded in mystery. You can’t directly bulk insert values using a query. However, you can bulk load using a tool like BCP, an API like SqlBulkCopy or via a query that tells SQL Server to read the rows out of a file and insert them (BULK INSERT or OPENROWSET(BULK…)). Then, if you use one of these means to do a bulk load and watch server activity using a Profiler trace or extended events, you’ll see an insert query but no row data. What’s going on?! Continue reading

T-SQL on the Wire: What takes place between client and server?

The answer to this question might shed light on some Microsoft® SQL Server® mysteries, might answer—or raise—security concerns and might help you administer, develop and debug better.

In this presentation, learn about Tabular Data Stream (TDS)—the protocol used for client-to-SQL Server interactions. In addition to a  high-level understanding of how TDS works, you’ll (hopefully!) come away with practical applications of this knowledge that should benefit you as a database administrator or developer.

 

Make It Easier for the DBA: Give SQL Connections the Application’s Name!

, , ,

Imagine you’re a database administrator tracing query activity on a server. You’d want all the clues you can get to help you figure out what’s going on. One clue is the name of the client program associated with each database session. This name, passed from client to Microsoft SQL Server when a connection is established, can be used by the DBA to differentiate between multiple applications running on a single client and to focus on (or filter out) activity from the same application across multiple clients.

Does your application pass a meaningful program name to the database server? Continue reading

Equating NULLs with EXISTS + INTERSECT

,

Can you rewrite WHERE Col1 = Col2 OR (Col1 IS NULL AND Col2 IS NULL) so that each column name is only used once without using the substitution method (that is, without doing something like WHERE ISNULL(Col1, '-2,147,483,648') = ISNULL(Col2, '-2,147,483,648'))?

WHERE EXISTS(SELECT Col1 INTERSECT SELECT Col2)

What…what?! That can’t be possible! I thought INTERSECT compares two result sets, not two columns from the same row of data. Continue reading

Swapping Data Sets Without Renaming Tables

Manipulating data directly in a table isn’t always practical. On occasion, performance requirements may dictate that the revised or replacement data set first be assembled in a separate table (a staging table) then switched in to replace the currently live data. Continue reading

For a Future Version: Re-Engineer Microsoft Access to Use a SQL Server Backend

,

At the core, both Microsoft Access and Microsoft SQL Server are relational database management systems (RDBMS), yet each uses a separate file format and code base. These differences add friction to the process of integrating the two technologies. Developers interacting with both back ends need to learn the idiosyncrasies of two systems. For administrators, the fact that each uses a different driver requires maintaining multiple driver configurations on systems that need to communicate with both. Power users are faced with the pain of figuring out how to cross this technology divide when growth dictates switching a home-grown Access application to a SQL Server backend. Continue reading

T-SQL Ordering Myth #2: ORDER BY in Views

,

Since a view is a saved query and queries can specify ordering, adding ORDER BY a view definition might seem a reasonable proposition. Try it and Microsoft SQL Server chokes. Then you learn the trick: include a TOP clause and SQL Server will be a-ok with ordering clauses in views. You want all rows returned so you add “TOP 100 PERCENT” to your definition and SQL Server is happy!

Yet you notice something strange. Sometimes the rows returned aren’t sorted according to the view’s ORDER BY clause. As you ponder this puzzlement, your mind wonders back to the roundabout syntax required to finagle ORDER BY into the view definition. Is something funny going on? But the answer eludes you. Continue reading

Quick Tip: Decrementing IDENTITY Column

Did you know that IDENTITY columns can count downward?! According to MSDN, IDENTITY’s second argument, named increment, “is the incremental value that is added to the identity value of the previous row that was loaded.” No constraint is given that increment must be positive (remember from Algebra that both positive and negative numbers can be added). Set increment to a negative number and IDENTITY will generate a descending sequence of numbers. Continue reading

T-SQL Ordering Myth #1: Clustered Indexes Determine Default Sort Order

,

If I query a table that has a clustered index without specifying an ORDER BY clause, the resultset will be sorted according to the clustered index” may sound reasonable. After all, since “a clustered index determines the physical order of data in a table” (per TechNet), isn’t it logical to assume that a query with no ORDER BY clause returns data sorted this way? Continue reading