Reducing data footprints on test systems


Picture the scene:

You are a DBA. You have to make everyone happy, including DEVs (yes, including them too). Part of that is to ensure that your DEVs have a fresh copy of your production data so that they can test their programming with current data and structures. The issue with that is your production database is multiple hundreds of GB in size and your test system doesn’t have the capacity to store a full copy of your database. Your IT Budget will also not stretch to purchasing more storage! What can you do?

I had a similar situation recently and took a look at the possibilties and here is what I came up with:

1. Try out tools like SQL Virtual Restore to allow you to “mount” a backup from a remote server avoiding the space issue all together. This is, of course, a solution that requires a license (no budget maybe?)

2. Use the great features offered to you by SQL Server Developer Edition: data compression to the rescue!

So of course, I chose door number 2.

As you may know, Developer Edition offers the same features as Enterprise Edition with the caveat that the instance cannot be used for production purposes in any way. This means that the seriously powerful and useful data compression becomes available for your test servers! This counts even if you only use Standard Edition in production – joy of joys! The good thing being that Developer Edition only costs around $50 (or less if you have an MSDN subscription) – even more joys!

If you have never seen/used compression take a quick look over on MSDN to see what it all is (Data Compression). Basically, you can compress data in indexes and tables at the storage level avoiding any query rewrites and still profiting from the storage savings. This can be a major advantage in terms of raw storage needs, but can also benefit you in high read environments with low RAM sizes. The data remains compressed when held in the buffer pool and is only de-compressed when being accessed. This means that you can keep more data in the buffer pool and reduce hard disk I/O (obviously not for free, compressing and de-compressing costs CPU cycles).  This may be acceptable on a test system with extremely constrained storage space.

The usage in my example scenario is now an even better proposition, not only can I reduce my storage footprint, I can also potentially increase test system performance on the I/O side of the equation (who doesn’t have I/O issues, especially on a test box!).

The next hurdle is of course identifying which indexes and tables that are in the database you want to squeeze down. This is possible via SSMS’ object explorer, but only if you want to spend an eternity doing so! The best way is to take a look at the meta-data tables/views/functions to interrogate the system objects. These are really interesting on their own (at least they are to me), but after writing my Index Script Creator, I realised the potential for using the information in these system objects to allow me to automate certain tasks like this one.

Similarly to the Index Script Creator, my Index Compressor Script (uploaded as a doc  but it is a .sql file really – stupid wordpress restrictions!) runs off and investigates all tables and indexes (down to the partition level) and then generates an index rebuild command to compress any currently non-compressed partitions of indexes / tables. The resulting commands can then be run and space can be reclaimed – just be aware that rebuilding indexes does cause log activity.

After compressing your tables and indexes you should have a much smaller amount of data (depending on how well your data compresses) and be able to reclaim some disk space.

I realise that this method will mean that you have to have a large enough disk to have the database in its uncompressed state to begin with, but you can end up with a cost effective solution to a budget-constrained test environment.

Happy zipping!

When COUNT() isn’t the only way to count


I have come across a situation a number of times in the past that seems to be one of those things that are so obvious when you see the solution, but can’t see them before the penny has dropped.

Imagine the following scenario:

You want to find the total number of orders that have the Order Status ‘A’ and the number of orders with an Order Status of ‘B’. This sounds like a simple enough request, that I’m sure you have heard of before.

Lets start off with some test data.

--Test Structure
USE master
go
IF DB_ID('Sandbox') IS NULL
BEGIN
    CREATE DATABASE Sandbox
END
GO

USE Sandbox
GO
IF OBJECT_ID('dbo.CountExample') IS NOT NULL
BEGIN
    DROP TABLE dbo.CountExample
END
GO
IF OBJECT_ID('dbo.Nums') IS NOT NULL
BEGIN
    DROP FUNCTION dbo.Nums
END
GO
-- Test Function to allow fast test data creation
CREATE FUNCTION [dbo].[Nums] (@m AS bigint)
RETURNS TABLE
AS
RETURN
WITH t0
AS (SELECT n = 1
UNION ALL
SELECT n = 1),
t1
AS (SELECT n = 1
FROM t0 AS a,
t0 AS b),
t2
AS (SELECT n = 1
FROM t1 AS a,
t1 AS b),
t3
AS (SELECT n = 1
FROM t2 AS a,
t2 AS b),
t4
AS (SELECT n = 1
FROM t3 AS a,
t3 AS b),
t5
AS (SELECT n = 1
FROM t4 AS a,
t4 AS b),
results
AS (SELECT ROW_NUMBER() OVER (ORDER BY n) AS n
FROM t5)
SELECT n
FROM results
WHERE n <= @m

GO
CREATE TABLE dbo.CountExample
(OrderId int NOT NULL,
OrderStatus char(1) NOT NULL)

GO

--Test data
INSERT INTO dbo.CountExample
(OrderId,
OrderStatus)
SELECT n,
CHAR(n % 27 + 64)
FROM dbo.Nums (1000) AS N
GO

Now that we have some test data and tables, we can take a look at what solutions are possible.

Solution 1:

The solution that I have seen come from a lot of people has been to basically run two queries, one for each Order Stautus and then collect these together returning the result.

Something along the lines of:

SELECT (SELECT COUNT(*) CountA
        FROM dbo.CountExample AS CE
        WHERE OrderStatus = 'A') CountA,
       (SELECT COUNT(*) CountB
        FROM dbo.CountExample AS CE
        WHERE OrderStatus = 'B') CountB

This delivers the correct result, but causes two separate queries to be run (one for each Order Status). There are variations of this solution, using sub-queries or CTEs, but I hope you get the idea that a separate COUNT() is required for each total that you want to calculate.

Solution 2:

The best way, that I know of, to achieve this would be to change the logic from a COUNT() to a SUM(). This sounds wrong at first, especially because the column Order Status is a char(1) and not an integer!

Take a look at how I have solved the problem with SUM():

 
SELECT SUM(CASE WHEN OrderStatus = 'A' THEN 1 ELSE 0 END) CountA,
       SUM(CASE WHEN OrderStatus = 'B' THEN 1 ELSE 0 END) CountB
FROM dbo.CountExample AS CE

Looking at the code, we can see that I have not just used SUM(), but also a CASE statement. CASE is one of my favourite constructs in T-SQL, as it allows you to perform logical processing of an entire set or only part of a set without filtering using a WHERE clause.

If you take a look at the execution plan, you will also see that the table is accessed once. This is an instant improvement over the “standard” solution of COUNT()-ing per Order Status and has the added bonus of never being noticeably more expensive, regardless of how many different Order Status totals are required.

So there you go.  COUNT() isn’t always the best way to count data in SQL Server.

SSMS ToolsPack – Powershell Turboboost


I have been using the rather brilliant SSMSToolsPack from Mladen Prajdić recently and love the query execution history feature.

I like to keep my history around for a while – I do a lot of work that then has to be repeated later – and this tool lets me not worry about forgetting to save that important query I ran for someone weeks ago.

I ran in to a little problem though.  All those query executions have to be stored somewhere, this is done by setting a folder to store the query text. 

SSMSToolsPack stores the query text in txt files, these can get out of control if you run enough queries.  They are then stored per day in a folder; so for today the queries would be stored in the folder “2010-12-03”.

If you want to search through the history and have a great number of files and folders, the search can get very slow (I suspect this has to do with the directory and file traversal).  I am lucky to have a small SSD on my main machine, I store source code and the execution history files/folders on there.  This speeds things up, but it seems that even then the search is sluggish (takes about 30 seconds to index on my machine).

I took a quick look at these files and saw that they were basically all the same content wise.  I tried just combining the files to see if that could improve the performance of searching and lo-and-behold search was blazingly fast!

Being lazy, I whipped up a script in Powershell to make this easier/semi-automatic and here it is:

clear host
$path = "" #Set path here!
foreach ($folder in Get-ChildItem $path)
{
  if ($folder.PSIsContainer -eq "False")
    {
      foreach ($file in Get-ChildItem $folder.FullName)
       {
         $target = $path+$folder.Name+"\output.txt" #set output file
         if ($file.Name -ne "output.txt")
          { #concatenate content of all files except output
            cat $file.FullName > $target 
            #del $file.FullName #delete the file after processing
          }
       }
    }
}

It is nothing special, but maybe if you use SSMSToolsPack and have experienced a similar slowdown, you can use this to help.

Presenting the Index Creator Script


Born of a need to originally script out indexes in a way similar to how SSMS creates them, I created a script to do just that. The original was very quick-n-dirty as can be seen here: http://ask.sqlservercentral.com/questions/16646/create-script-for-indexes
I had not accounted for very much, other than the indexes as they were. All options, schema information etc. was ignored, as we have nothing special at work and I really needed those indexes quickly.

Since posting on ASK, I have tinkered on-and-off with the script for a while. I am now at a point where I think other people could really profit from it and no longer have it stuck to some hard-coded schemas etc.

Presenting the Index Creator Script v1.0! This script will go through the current database, finding all indexes (optionally system indexes too) and supply you with create index scripts.

It is clever enough to spot the difference between / usage of :

– Primary Keys
– Unique Constraints
– Clustered and Non-Clustered Indexes
– Filtered Indexes (produces the filter too)
– Included Columns
– Partitioned Tables/Indexes (although the partition schema and functions are not produced – yet!)
– Data Compression (on a partition level if used – yes this is possible!)
– Fill Factor
– Index Padding
– Locking (Row and Page)

I have supplied two versions of the code; one for SQL 2005 and one for SQL 2008 and above. This is done as SQL 2008 offers Data Compression, which is implemented in the indexes and partitions. Some of the script relies on this information and would not be backwards compatible.

It has been very interesting coding this script, as it has enlightened me on the structures in SQL server with regards to indexes. For example, as of SQL 2005, regardless of edition, SQL Server creates indexes using partitions. Although partitions cannot be used by editions lower than Enterprise/Developer Edition, all indexes are created with at least one partition. This makes sense, as that would mean there would have to be a separate structures depending upon edition. This way, regardless of edition, the storage engine works the same, you just don’t get the option of creating partitions on an edition lower than Enterprise/Developer. As soon as you migrate a database to Enterprise Edition, you get the possibility of then splitting the indexes on to multiple partitions.

Even better than that, I found out that indexes can be compressed by partition. I sort of knew this already, but in writing the script I saw this in even more clarity. Each partition of an index can use a different level of compression. This can be very interesting, especially if the costs of compression are high, but the benefits in storage are high too. Think of a CPU bound system where some partitions are accessed often and would need a lower compression to reduce CPU load, with other partitions that are accessed rarely which can benefit from the higher compression ratio.

I hope these scripts are of some benefit. If you have comments/questions/suggestions, please get in touch.

Index Creator Script – This is a zip file. Download, change file extension and open in your favourite ZIP manager (damn you wordpress!). There are 2 .sql files in there (one for 2005 and one for 2008). Disclaimer – use at your own risk, I am not responsible if it breaks your PC/Server.

In reality, this script can’t break things, but you have been warned!

UPDATE: Thanks to @Fatherjack for the quick heads-up on a syntax error. Things should look good now though! 🙂

Let technology replace your ageing brain


After seeing a question on ASK (Sqlsercentral version of Stackoverflow) asking for help with code to extract index meta data, I took a look into my little box of coding tricks.  I had put something together a while back to rename indexes to fit a certain naming scheme that almost fit the job.  I promptly posted my answer and the OP was suprised at the speed of the reply.

This proved to me again, that keeping all scripts that you ever create is really important.  If you have written it, save it somewhere permanent.  Ideally you will keep these scripts on a network share or on the web, so that you can access it any time, anywhere.  I have learned the hard way, that the little innocent script you wrote and threw away, is going to be needed again.  This normally happens about a month or so later, and tools like SSMS Toolpack with the excellent Query Execution History can help, but not as good as a script collection.

I know that I will be updating my script collection and will post the scripts and a little note here as and when I get the time.

So remember, save your scripts and be prepared!