Press "Enter" to skip to content

Author: Kevin Feasel

Slow Merge Replication Initialization

Published 2019-12-16 by Kevin Feasel

Gonzalo Bissio covers one reason why initializing merge replication might be slow:

Since 1 week (Yes… 1 week) they were waiting for the replication re-initialization to finish since they introduced some changes on the system and they needed to reinitialize the replication again. The rate of the records between the publisher and the subscribers were too low (20 records per second). Since this application is used globally they write records on all of the regions (then the merge replication manage them).

Read on to see what Gonzalo’s company had going on and how they fixed it.

Comments closed

YARN Container Sizing

Published 2019-12-13 by Kevin Feasel

Dmitry Tolpeko explains why large YARN containers aren’t always a great idea:

First I noticed that the job used only 100 containers i.e. just one container per cluster node. This was very suspicious as Hive uses the Apache Tez execution engine that can run concurrently only one task in a container.

Looking at the Hive script I found:

set hive.tez.container.size = 10240; -- 10 GB

Looks like someone had a memory problem with this query before and wanted to solve it once and forever!

Read on to see why this was not a great idea.

Comments closed

Geospatial Data Processing with Databricks

Published 2019-12-13 by Kevin Feasel

Razavi and Michael Johns walk us through examples of processing geospatial data with Databricks:

Earlier, we loaded our base data into a DataFrame. Now we need to turn the latitude/longitude attributes into point geometries. To accomplish this, we will use UDFs to perform operations on DataFrames in a distributed fashion. Please refer to the provided notebooks at the end of the blog for details on adding these frameworks to a cluster and the initialization calls to register UDFs and UDTs. For starters, we have added GeoMesa to our cluster, a framework especially adept at handling vector data. For ingestion, we are mainly leveraging its integration of JTS with Spark SQL which allows us to easily convert to and use registered JTS geometry classes. We will be using the function st_makePoint that given a latitude and longitude create a Point geometry object. Since the function is a UDF, we can apply it to columns directly.

Looks like they have some pretty good functionality here, and they have shared the demos in notebook form.

Comments closed

Fun with NULL

Published 2019-12-13 by Kevin Feasel

Itzik Ben-Gan takes us through some of the complexities of NULL:

NULL handling is one of the trickier aspects of data modeling and data manipulation with SQL. Let’s start with the fact that an attempt to explain exactly what a NULL is is not trivial in and of itself. Even among people who do have a good grasp of relational theory and SQL, you will hear very strong opinions both in favor and against using NULLs in your database. Like them or not, as a database practitioner you often have to deal with them, and given that NULLs do add complexity to your SQL code writing, it’s a good idea to make it a priority to understand them well. This way you can avoid unnecessary bugs and pitfalls.

This article is the first in a series about NULL complexities. I start with coverage of what NULLs are and how they behave in comparisons. I then cover NULL treatment inconsistencies in different language elements. Finally, I cover missing standard features related to NULL handling in T-SQL and suggest alternatives that are available in T-SQL.

This is definitely worth the read.

Comments closed

Power BI and Azure Synapse Analytics

Published 2019-12-13 by Kevin Feasel

James Serra gives us some insights on the future of Power BI and how it relates with Azure Synapse Analytics today:

As an example of the speed of each layer, during an Ignite session (view here), there was a Power BI query run against 26 billion rows that was returning a sum of store sales by year. The same query was run three times using a different layer:

1. Using a DirectQuery against tables in SQL DW took 8 seconds
2. Using a DirectQuery against a materialized view in SQL DW took 2.4 seconds.  Note you don’t have to specify that you are using a materialized view in the query, as the SQL DW optimizer will know if it can use it or not
3. Using a Aggregation table that is Imported into Power BI took 0 milliseconds

Keep in mind this is all hidden from user – they just create the report.  If they do a query against a table not in memory in Power BI, it will do a DirectQuery against the data source which could take a while.  However, due to SQL DW result-set caching, repeat DirectQuery’s can be very fast (in the Ignite session they demo’d a DirectQuery that took 42 seconds the first time the query was run, and just 154 milliseconds the second time the query was run that used result-set caching).

There’s some interesting information in here, especially around Power BI eventually taking over Azure Analysis Services’ space in the market.

Comments closed

Azure Data Factory Triggers

Published 2019-12-13 by Kevin Feasel

Cathrine Wilhelmsen continues a series on Azure Data Factory by looking at triggers:

One important thing to note is that all times are in UTC. And since UTC does not observe daylight saving time… Well, let’s just say that if you need to execute pipelines during the workday and you have business users waiting for data, you may want to plan some trigger maintenance on the days when you fall back or spring forward. I know. Ugh 🙂 I’m hoping for better timezone support in the future 🤞🏻

Schedule triggers and pipelines have a many-to-many relationship. That means that one schedule trigger can execute many pipelines, and one pipeline can be executed by many schedule triggers.

Time-based triggers aren’t the only options, however—Cathrine also looks at the other three possibilities.

Comments closed

Posting SQL Query Results to Teams with Powershell

Published 2019-12-13 by Kevin Feasel

Michael Bourgon shows how we can post SQL Server result sets to Microsoft Teams channels:

So…. you want to post to a Teams channel automagically.  Should be simple, and it is!  Alas, it means you have to ignore most of the documentation.  Let’s do this!

1) Here’s how to hook it up with your channel.  Note that when I created a brand new “Team”, it took about 5-10 minutes before I was able to add the webhook connector – prior to that, I got a “channel does not exist or has been deleted”.

Michael takes us through it step by step and also includes things you should avoid, including misleading documentation.

Comments closed

Flink 1.8.3 Released

Published 2019-12-12 by Kevin Feasel

Hequn Cheng announces Flink 1.8.3:

The Apache Flink community released the third bugfix version of the Apache Flink 1.8 series.

This release includes 45 fixes and minor improvements for Flink 1.8.2. The list below includes a detailed list of all fixes and improvements.

We highly recommend all users to upgrade to Flink 1.8.3.

There’s a nice list of bugfixes in the update.

Comments closed

Kryo Serialization in Spark

Published 2019-12-12 by Kevin Feasel

Pinku Swargiary shows us how to configure Spark to use Kryo serialization:

If you need a performance boost and also need to reduce memory usage, Kryo is definitely for you. The join operations and the grouping operations are where serialization has an impact on and they usually have data shuffling. Now lesser the amount of data to be shuffled, the faster will be the operation.
Caching also have an impact when caching to disk or when data is spilled over from memory to disk.

Also, if we look at the size metrics below for both Java and Kryo, we can see the difference.

Sounds like it’s better overall but requires some custom configuration.

Comments closed

Finding the Max Value Across Multiple Columns

Published 2019-12-12 by Kevin Feasel

Erik Darling shows a couple techniques for finding the maximum value across several columns, whether they’re in one table or in more than one:

It’s sorta kinda pretty crazy when every major database platform has something implemented, and SQL Server doesn’t.

Geez, even MySQL.

But a fairly common need in databases is to find the max value from two columns.

Maybe even across two tables.

Read on to see how you can do this.

Comments closed