Press "Enter" to skip to content

Author: Kevin Feasel

Getting Around in Spark

Tomaz Kastrun continues a series on Apache Spark. Day 3 shows off the CLI and web UI:

In CLI we will type and run a simple Scala script and observe the behaviour in the WEB UI.

We will read the text file into RDD (Resilient Distributed Dataset). Spark engine resides on location:

/usr/local/Cellar/apache-spark/3.2.0 for MacOS and
C:\SparkApp\spark-3.2.0-bin-hadoop3.2 for Windows (based on the blogpost from Dec.1)

Day 4 compares local mode versus cluster mode:

Finding the best way to write Spark will be dependent of the language flavour. As we have mentioned, Spark runs both on Windows and Mac OS or Linux (both UNIX-like systems). And you will need Java installed to run the clusters. Spark runs on Java 8/11, Scala 2.12, Python 2.7+/3.4+ and R 3.1+. And the language flavour can also determine which IDE will be used.

Day 5 shows the setup of a Spark cluster:

Spark can run both by itself, or over several existing cluster managers. It currently provides several options for deployment. If you decide to use Hadoop and YARN, there is usually the installation needed to install everything on nodes. Installing Java, JavaJDK, Hadoop and setting all the needed configuration. This installation is preferred when installing several nodes. A good example and explanation is available here. you will also be installing HDFS that comes with Hadoop.

Check out all three posts and get caught up on Spark.

Comments closed

Indexing and Window Functions

I continue a series on window functions in SQL Server:

If you’ve been around the block with window functions, you’ve probably heard of the POC indexing strategy: Partition by, Order by, Covering. In other words, with a query, focus on the columns in the PARTITION BY clause (in order!), then the ORDER BY clause (again, in order!), and finally other columns in the SELECT clause to make the index covering (not in order! though it doesn’t hurt!).

But do read on to understand why this is not sufficient.

Comments closed

Troubleshooting Timeouts with Import Refresh

Chris Webb begins a series on troubleshooting timeouts:

If you’re working with a large Power BI dataset and/or a slow data source in Import mode it can be very frustrating to run into timeout errors after you have already waited a long time for a refresh to finish. There are a number of different types of timeout that you might run into, and in this series I’ll look at a few of them and discuss some of the ways you can work around them.

In this post I’ll look at one of the most commonly-encountered timeouts: the limit on the maximum length of time an Import mode dataset refresh can take. 

Click through to see the limits and ways to (sort of) get around them.

Comments closed

Tracking Value Changes in SQL Server

Tomas Zika wants to track a change:

This time a colleague from work asked how to best find a culprit that has been changing a specific cell in a table. It could be an automated process, application logic, application user or even an ad-hoc statement – we didn’t know. The table has many different access patterns, some of which are frequent. Ideally, we don’t want to monitor everything and sift through it.

We wanted to learn the who, the how and then ask why? If you like to know the whole journey, read on. Otherwise, you can skip to section Eureka moment

Click through for the Eureka moment. It is important to embrace the power of “and.”

Comments closed

The User-Assigned Managed Identity in ADF

Asanka Padmakumara takes a look at the user defined managed identity:

If you are familiar with Managed Identity concepts in ADF, each ADF instance comes with own System Assigned Managed Identity (MI). We can use that MI to control ADF’s access to any data sources which support Azure AD based authentication. This is considered to be the most secured and recommended way of authenticating ADF with cloud systems. If not, you can use Azure Key vault to store credentials. Let’s take an example on to discuss how User Assigned Managed Identity helps for manage access within multiple ADF environment.

Click through to see how the user assigned managed identity makes life better.

Comments closed

SQL Server Backend for Django

Warren Chu announces a new version of the SQL Server 3rd Party Backend for Django:

We have released version 1.1 of the SQL Server 3rd Party Backend for Django. This release contains support for the upcoming release of Django 4.0, as well as a number of issue fixes.

Our plan is to time releases to coincide with major releases of Django and SQL Server, to ensure users of this project can keep up to date with Django while continuing to use SQL Server as a backend.

Read on to see what this entails.

Comments closed

Choosing a Statistical Test

Antoine Soetewey has a handy chart for us:

Being a teaching assistant in statistics for students with diverse backgrounds, I have the chance to see what is globally not well understood by students.

I have realized that it is usually not a problem for students to do a specific statistical test when they are told which one to use (as long as they have good resources and they have been attentive during classes, of course). However, it appears that the task is much more difficult for them when they need to choose what test to do.

Click through for the chart, as well as a PDF version. H/T R-Bloggers.

Comments closed

Installing Apache Spark

Tomaz Kastrun continues a series on Apache Spark:

Installing Apache Spark on Windows computer will require preinstalled Java JDK (Java Development Kit). Java 8 or later version, with current version 17. On Oracle website, download the Java and install it on your system. Easiest way is to download the x64 MSI Installer. Install the file and follow the instructions. Installer will create a folder like “C:\Program Files\Java\jdk-17.0.1”.

Read on for instructions for both Windows and MacOS. You can also create a container running Spark, which is another helpful method.

Comments closed

Locking Issue with Columnstore Indexes

Joe Obbish troubleshoots an issue on tables with columnstore indexes:

I recently ran into a production issue where a SELECT query that referenced a NOLOCK-hinted table was hitting a 30 second query timeout. Query store wait stats suggested that the issue was blocking on a table with a nonclustered columnstore index (NCCI). This was quite unexpected to me and I was eventually able to produce a reproduction of the issue. I believe this to be a bug in SQL Server that’s present in both RTM and the current CU as of this blog post (CU14). The issue also impacts CCIs as well but I did significantly less testing with that index type.

Read on for the issue, how you can replicate it, and a couple ways to work around it.

Comments closed

Why the Optimizer Doesn’t Look at Buffer Pool Data

Paul Randal has an explanation for us:

SQL Server has a cost-based optimizer that uses knowledge about the various tables involved in a query to produce what it decides is the most optimal plan in the time available to it during compilation. This knowledge includes whatever indexes exist and their sizes and whatever column statistics exist. Part of what goes into finding the optimal query plan is trying to minimize the number of physical reads needed during plan execution.

One thing I’ve been asked a few times is why the optimizer doesn’t consider what’s in the SQL Server buffer pool when compiling a query plan, as surely that could make a query execute faster. In this post, I’ll explain why.

This is an interesting post because it explains why the developers of the database engine would purposefully ignore something that could make things faster, but at a potentially devastating cost.

Comments closed