Press "Enter" to skip to content

Curated SQL Posts

Include Filtered Index Columns

Published 2015-12-08 by Kevin Feasel

Erik Darling shows how to make filtered indexes a bit more robust:

But why oh why didn’t SQL use my filtered indexes for even smaller subsets of the filter condition? It seemed insane to me that SQL would know the filter for the index is on (x > y), but wouldn’t use them even if (z > x).

The solution was to put the filtered column in the include list. This lets SQL generate statistics on the column, and much like getting rid of the predicate key lookup, allows you to search within the filtered index subset for even more specific information.

Filtered indexes are as useful as they are mercurial.

Comments closed

Cluster Name Changes

Published 2015-12-08 by Kevin Feasel

Matt Slocum ran into a production issue with a cluster name that changed:

Next is just a simple matter of updating the registry value.  I launched Registry Editor (RegEdit.exe) and navigated to HKLM > Software > Microsoft > Microsoft SQL Server > MSSQL12.MSSQLSERVER > Cluster
Note: MSSQL12.MSSQLSERVER will vary based on SQL version (SQL 2012 is “MSSQL11”) and instance name (“MSSQLSERVER” is the default instance and named instances are the actual instance name instead of “MSSQLSERVER”).

Within the Cluster key I updated the value of ClusterName to match what was specified in Failover Cluster Manager

This is a good diagnosis and resolution of an issue.

Comments closed

Fitbit BI

Published 2015-12-08 by Kevin Feasel

Reza Rad has a three-part series on applying BI tools (specifically, Power BI) to Fitbit.

Part 1:

So for this post we are going to build that dashboard (not all of that obviously, because we don’t have the data required for all of that), but most part of it with Power BI. You will see how easy and powerful is Power BI in this kind of scenarios, and you will see how you can be the BI Developer of Fitbit in a few steps of building this demo.

Part 2:

Unfortunately Power Query or let’s say Power BI doesn’t have a loop structure, and that is because of the functional structure of this language. However there are data structures such as Table and List that can be easily used with each singleton function to work exactly as a loop structure does. Here in this post I will get you through the process of looping into files in a directory and processing them all, and finally combining them into a large big table. You will also learn some Power Query M functions through this process.

Part 3:

Fitbit calculates based on my current weight and age (I assume) how much calories I have to spend each day. I don’t know that calculation, So I create a static measure with the value of 2989 for the amount of calories I have to spend each day. I also create StepsCap measure with 12000 value showing that I have to walk 12000 steps a day, and another one for FloorCap with the value of 10. I created a Calories HighEnd measure with 5000 calories as value (I will die if I burn more than that!). You can create all these measures easily in Data tab.

This is a nice combination of work and play, building an interesting system with a data set interesting to the author and freely available.

Comments closed

Tuning With BlitzIndex

Published 2015-12-08 by Kevin Feasel

Koen Verbeeck on using sp_BlitzIndex:

What’s great is that this script also provides you with the URLs to knowledge articles on the Brent Ozar website. If you don’t understand one of the results, you can immediately look it up and read about it.

By focusing on the results of sp_BlitzIndex script, I could boost performance in just a few hours of work. This near real-time data warehouse is the source for a reporting application used by dozens of people in the field, and you could immediately tell it worked a lot faster. Awesomesauce.

There are several interesting scripts in the suite; check them out if you’ve not already.

Comments closed

More On Server 2016 Pricing

Published 2015-12-08 by Kevin Feasel

Denny Cherry thinks Windows Server 2016 pricing changes won’t matter that much:

So looking at this chart we can see that any machine with less than 8 cores per socket and two sockets or less will cost you exactly the same as it will today.  Also any machine with one socket and less than 16 cores will also cost the same as it costs today.

I’ll admit that I don’t know much about licensing, but rumblings at my company are that our server licensing costs are going to go up significantly.  As always, check with one of the four people on Earth who understand Microsoft licensing.

Comments closed

Select Failures

Published 2015-12-08 by Kevin Feasel

Paul Randal walks us through a complicated scenario involving corruption:

An interesting corruption problem cropped up on the MCM distribution list yesterday and after I figured it out, I thought it would make a good blog post in case anyone hits a similar problem.

In a nutshell, the problem was corruption such that a simple SELECT * query failed, but a SELECT * query with an ORDER BY clause worked.

This is interesting, but hopefully you don’t encounter these types of problems very often in your environment.

Comments closed

Partitioning On The Cheap

Published 2015-12-07 by Kevin Feasel

Aaron Bertrand shows us how to partition on the cheap:

The TL;DR of this is that you can use filtered indexes to keep all of your “hot data” in a separate physical structure, and even on separate underlying hardware (you may have a fast SSD or PCIe drive available, but it can’t hold the whole table).

Using filtered indexes is an interesting way of solving issues that partitioning normally would help you out with.  I’m looking forward to the next part in the series.

Comments closed

Actual Number Of Rows Read

Published 2015-12-07 by Kevin Feasel

Kendra Little shows us something new, the Number of Rows Read tag in an execution plan:

Execution plans got a cool new piece of diagnostic information in SQL Server 2012 SP3, and we believe we’ll soon have this change in SQL Server 2014 and the coming SQL Server 2016: “Number of Rows Read”.

In fancy language, this is “better diagnostics” when a query plan has “residual predicate pushdown” (KB 3107397).

In human language, SQL Server will now tell you “How many rows did I really have to read, even if I have a hidden filter in here?”

This appears in actual execution plans only. Sorry, there is no such thing as “Estimated Number of Rows Read” that I can find.

This is another piece of useful information now available in execution plans.  Hopefully it’ll be supported in SQL Server 2014 and 2016 soon enough.

Comments closed

Naming Azure Servers

Published 2015-12-07 by Kevin Feasel

With Azure SQL Database, you can choose the server name:

Once you select the option to create a new server, you see similar options to those you saw in the Azure Management Portal. However, this time you see a server name.

Alas, you can again name your own database servers!  As you can see above in the Azure Preview Portal, the server has been created with the provided name. Then if we switch over to the Azure Management Portal, we will see the same.

Choose your names wisely.

Comments closed

Removing Bad Execution Plans

Published 2015-12-07 by Kevin Feasel

Andrea Allred shows one way of removing a bad query plan:

If you click on the query_plan link, you can see what the plan looks like.  After you have reviewed it and determined the plan is bad then you can paste your plan handle over the one below to remove it from the proc cache.

DBCC FREEPROCCACHE normally is something you don’t want to play with in production, but this is narrowly focused enough not to harm you down the line.

Comments closed