Wednesday, February 9, 2011

And you may say to yourself… MY GOD! WHAT HAVE I DONE?

Have you ever found yourself living David Byrne’s dire prophecy? ( Don’t try to reenact…  the 80’s were more tolerant to bizarre body movements, tics and twitches advertised as a ‘dance’)
Specifically, when you develop SQL, has it ever happened to you that you’ve casually run some UPDATE or DELETE statement,  and got the message that… say, way more records than you had intended were just deleted? And you sit there, staring at the screen, letting the fact of what just happened enter your mind little by little… and the only thought in your mind is… “holy sh**!”
Hopefully you’ve got backup. And a pretty recent one too. And the time to go through the restore hassle.
SQL doesn’t have an ‘undo’. That bothered us for years. We have partially answered this problem with our Data Profiler, which enables you to take quick snapshots of the data before you’re changing anything. But that’s not a full solution.
We’ve now taken a further step in that direction:  Our new Colombo can take a DML statement (INSERT, UPDATE or DELETE) detect exactly what changes its going to make, and generate the SQL script that will do the exact opposite of what the statement does. NICE. Well, we think so. You simply paste your statement into Colombo, select your database, and that’s it!
There is still considerable way to walk to provide a full undo to every sql operation, batch script, and stored procedure that you might call. And we intend to walk that road. For now, we’ve taken the first step. Think we’re crazy? Let us know…
(Now that song will be stuck in my head for the next 3 days. It’s the same problem with anything 80’s….)

Sunday, February 6, 2011

Useful information we can get out of a query execution plan

Much has been written about Query Execution Plans (we’ll call them QEPs), and how they can be used for analyzing how SQL Server interprets your TSQL code and what it does behind the scenes – information you can use to do such things as find bottlenecks, build indexes where they might help, and generally improve the performance of your code.
Well, we have found more uses for it. There is more information in the QEP, information that’s not typically used, yet is highly useful and very easy to retrieve:

1.     Is your code broken?

TSQL code can get ‘broken’ all the time. You might be referencing tables or columns that don’t exist. They might have even existed at the time you wrote the procedure, but have been removed since, leaving your procedure ‘broken’ -  It will raise an error next time its executed. How many of those do YOU have in your database?
Now, if a code is broken, SQL Server will let you know about that when attempting to retrieve its execution plan. Strictly speaking, this is not QEP information. Its information you will get while trying to get the QEP. And that is not the only way to get that information. But it’s the easiest.

2.     What entities are used in for what operation?

Among all the logical and physical operations, and effects on memory etc, QEP also contains the very basic information of what entities it is working with, and what it does on such entities. This information can be then used to build a basic ‘browser database’, which you can use to answer questions like: which code inserts to this table? I got data missing… which procedure removes records from here?

In order to get any SQL’s execution plan in SQL Server, all you need to do is:


write your SQL here


If you get an error, this means the stored procedure is broken, and the error would say why. Otherwise, you can get the execution plan in a tabular format, and retrieve the specific entities used from it.

We put this functionality into one of our freebie products, Diana Lite. You can build the full ‘browser database’ out of the collection of any database’s SQL Code, then easily filter it and get to whatever you are looking for:

Sweet, Huh?