A few of my favorite blogs

Today I decided to give credit and a special thank you to a lot of people in the SQL community whom have either helped me directly or indirectly with their knowledge and skills and most importantly their willingness to share back with the community.

So who do I look to for expert reliable sources for all things SQL Server. Let me start with my favorite trainers who blog. Without advanced training I would not have the skill set I have today. You can learn a lot on the job and my first DBA job lasted 6 years, covered 5 different database technologies across a few hundred servers. You can learn how to firefight well in that environment, but you rarely get a chance to hone your skills in one area. That is where training comes in. It is easy to get caught up in the day to day tasks of your particular job, and loose sight of what is new and changing. SQL Server has a lot of features that may not be applicable to your current job, but important to other jobs you may want to pursue in the future.

My first great training was Kalen Delaney, in Portland OR a few years ago, shortly after SQL 2005 had come out. The class blew my mind. Literally, I went in to the class feeling pretty competent in my skills and came out feeling very humble, and with a huge list of things I wanted to fix, change, and improve in my current job. In addition it made me want to know more and keep up with the changing technology. For that I thank Kalen and refer to her blog often as a great resource!

Kalen Delaney
http://sqlblog.com/blogs/kalen_delaney/default.aspx

As a side note I had the privilege of taking her internals class a second time when SQL 2008 first came out, it wasn't in person, but still was excellent, and I highly recommend her trainings.


Last year was the first time I had the pleasure and opportunity to take a course from Kimberly Tripp and Paul Randal. Wow, what a fantastic training. By the time I took their class I had been a production DBA for 6 years with lots of training, and was supporting over 2000 databases on more than 170 servers. I felt I had a pretty good understanding of all things SQL. My first surprise was that their was no computers or labs, it was more reminiscent of a graduate school class, with lecture, and discussion and some demonstration. It was fantastic, every day you felt like your brain had just ran a marathon. I learned a lot and tuned what I already knew even more. The networking was fantastic as well. I am very excited I get to go back for the second Immersion Event this August.

Kimberly Tripp
http://www.sqlskills.com/blogs/kimberly/

Paul Randal
http://www.sqlskills.com/blogs/paul/


Ok, I haven't taken a class from Brent, but I hear that the SQL Cruise is pretty awesome. I have met him once and he's a really nice guy. His social influence in the SQL Server realm is incredible and I have used his blog on many occasions.

Brent Ozar
http://www.brentozar.com/


I've been to a couple of Buck's presentations and he is a fantastic presenter, his blog is great too. Lots of good DBA administration stuff there, great ideas. In fact, one of Buck's talks got me started on tracking performance baselines for all of my systems.

Buck Woody
http://blogs.msdn.com/b/buckwoody/


In the category of I've never met, but am very grateful that they share their knowledge with the community, here are a few of my favorites. There are others too, but I use Glen and Ola's scripts all the time. Fantastic stuff.

Ola Hallengren
http://ola.hallengren.com/

Glen Berry
http://sqlserverperformance.wordpress.com/

Erland Sommarskog
http://www.sommarskog.se/index.html

Michelle Ufford
http://sqlfool.com/

More fun with Management Data Warehouse

Management Data Warehouse (MDW) really is a cool feature that was added to SQL server in 2008. While there are some issues, and the default out of the box setup might kill you on disk space and activity if you have a lot of instances sharing the same MDW, overall it is a great resource with tons of potential.

Similar to the canned reports that come with SQL Server Management Studio for database and instance details, there are canned reports that come with the setup of the MDW.

The Management Data Warehouse Overview report has three columns with links to the reports for Server Activity, Query Statistics, Disk Usage for each server you have collectors defined for.

A closer look at the Query Statistics History report gives a great overview of the top 10 queries for CPU, Duration, Total I/O, Physical Reads, and Logical Writes. You can adjust the slider at the top for a particular time range you are interested as well.

These reports are great for looking at performance over time and identifying queries that could be improved. However, I wanted to know more information and see where in the MDW the data for this graph is stored. It turns out there is a very handy stored procedure in the MDW called. [snapshots].[rpt_top_query_stats] that queries the snapshots.query_stats, core.snapshots, and snapshots.notable_query_text to get some very useful data.

The procedure takes an instance name, start time, end time, time window size, time interval (in minutes), the order by criteria (CPU,Physical Reads,Logical Writes, I/O, and Duration) and Database name.

So rather than write my own code to do the same thing I just created a new copy of this stored procedure and tweaked it a bit. instead of top 10 I changed it to be top 100. Next I tweaked some of the parameters to make it so I could get a weekly snapshot of top 100 queries by CPU, I/O etc. Now I have my own [snapshots].[rpt_top100_query_stats] stored procedure.

Next, of course I created a new SSRS report that calls my new procedure and can send me a nice pretty email of the data on whatever schedule I choose with whatever metrics I choose.

SSRS 2008 R2 is really cool, if you haven't figured out how powerful it is, take the time to play with it a bit. There are so many great things about it and I use it on a regular basis in my position as a DBA. I know I don't use it to it's full potential, I'll leave that to the BI experts. However, having numerous metric reports is incredibly helpful to me in my job. I use SSRS for monitoring trends over time and have also used it to build reports for developers to see how their code is actually preforming, plus many more.

Creating Custom Data Collectors SQL Server 2008 R2 Part 2 SSRS

In Part 1 I discussed creating a custom data collector based off of one of the SQL Server Management Studio (SSMS) canned reports. For Part 2 I will create a SQL Server Reporting Service (SSRS) report that runs off of the Management Data Warehouse (MDW) and queries my custom collector data to build a nice graph of my top tables.

The first thing I needed was a way to connect to my MDW from my reporting server instance. Since I wanted to set up my report to be an email subscription, which means I had to have stored credentials on my SSRS server. I created a new custom sql login on my SQL Server instance where my MDW database lived. I gave it read-only or db_datareader role on that database and then verified it and then started working on the queries I wanted to run.

In the previous post I created a new collector that created a table with stats for all the tables in my database. If I look in my MDW database I now have a new table under the custom_snapshots schema with data from my new collector.

The table now has the basic stats for the database tables being monitored. Including Schema,tablename,row_count,reserved,data,index_size, unused, the collection_time and snapshot_id (this gets incremented every time the collector runs so you can distinctly identify collections).


The following query will give me a snapshot of the top ten tables as of the last collection time.

select top 10 tablename,row_count FROM [MDW].[custom_snapshots].[table_stats] Where databasename='AdventureWorksDW2008' AND snapshot_id = (select MAX(snapshot_id) from [MDW].[custom_snapshots].[table_stats] ) order by row_count desc

Switching gears I can now open up BIDS and create a new reporting services project. You can use the wizard if you so choose or add things manually, I won't show that here but there are plenty of sites on how to get started with SSRS. For this example I want a chart, see here for a good tutorial Tutorial: Adding a Bar Chart to a Report (Report Designer).

Next in Business Intelligence Studio (BIDS) create a new project for your MDW Reports.

Whether using the wizard or creating datasource, datasets, and reports separately the key is that I will need all three for this example. (hint: the wizard will step you through creating these three things in case you get confused).
For my datasource I want to use the readonly account I created and granted access to my MDW database earlier.

Next I create a new dataset against that datasource with my query see above.

Now I need a new report (in this case I did not use the wizard). Once I have my report area visible in design view I can add a chart. I chose Bar chart for this example.

Fill in the chart data with the appropriate columns from your data set.

I added enable scale breaks to the horizontal axis since a few tables are much larger than others so it looks nicer.

Lastly I preview the report to make sure it looks right.

Now I can deploy it to my SSRS server if I want. Once there I can setup a subscription to email the report to whomever I want on whatever schedule I want. For my final report in my case I wanted a weekly report that showed table growth over time. So I had to adjust the query and chart accordingly. It now looks like this one except or course mine has numbers and labels on it.