Press "Enter" to skip to content

Category: ETL / ELT

SSIS Design Preferences

Meagan Longoria systematizes a set of preferences regarding Integration Services package and ETL process design:

– Every table should have InsertDateTime and UpdateDateTime columns. The UpdateDateTime column should be populated with the same value as the InsertDateTime column upon creation of the row, rather than being left null.
– Whatever you use to create tables, include primary keys, foreign keys, and indexes with your table definitions. Provide explicit constraint names to simplify database comparisons. You can disable your foreign keys, but they need to be there to provide that metadata.
– Separate your final dimensional/reporting tables from audit tables and staging tables. This can be done with separate schemas or even separate databases.

People have added some more thoughts in the comments as well.

Comments closed

Dimensional Load with Databricks

Leo Furlong shows how we can load an Azure SQL Data Warehouse dimension with Databricks:

Ingesting data into the Data Lake occurs in steps 1 and 2 in our architecture.  Azure Data Factory (ADF) provides an excellent mechanism for loading data from source applications into a Data Lake stored in Azure Data Lake Store Gen2.  In fact, Microsoft offers a template in the ADF Template gallery which provides a metadata driven approach for doing so.  The template comes with a control table example in a SQL Server Database, a data source dataset and a data destination dataset.  More on this template can be found here in the official documentation.

I appreciate that this is a full walkthrough of the process, not just one step.

Comments closed

Testing ETL Pipelines

Ed Elliott has started a new series on testing ETL pipelines:

We test in production, this means we have monitoring and do things like have phased roll-outs using feature flags, or we roll-out to select customers first, prove it then roll it out to everyone else. Testing in production doesn’t mean hacking around getting some process to work. We don’t test “on production” (hacking), we test “in production” – while we are in production we are continually testing, and if anything goes wrong, we have alerts and can deal with it.

Testing pipelines feels difficult because there are so many moving pieces, but if you design for testability (e.g., being able to tee off samples of data, send test records through, etc.), things get easier.

Comments closed

Don’t Truncate Facts and Dimensions when Loading Data

Meagan Longoria explains why a truncate-and-reload strategy for data warehouses isn’t a good look:

Every once in a while, I come across a data warehouse where the data load uses a full truncate and reload pattern to populate a fact or dimension. While it may not be the end of the world for a small table, it does concern me and I usually recommend to redesign the load. My thoughts below on why this is an anti-pattern are true for using the actual TRUNCATE TABLE statement as well as executing a DELETE statement with no WHERE clause.

Read on for some great advice, including an exception to the rule.

Comments closed

PolyBase in SQL Server 2019

Ben Weissman takes us through SQL Server 2019’s PolyBase enhancements:

Isn’t that the same thing, as a linked server?
At first sight, it sure looks like it. But there are a couple of differences. Linked Servers are instance scoped, whereas PolyBase is database scoped, which also means that PolyBase will automatically work across availability groups. Linked Servers use OLEDB providers, while PolyBase uses ODBC. There are a couple more, like the fact that PolyBase doesn’t support integrated security, but the most significant difference from a performance perspective is PolyBase’s capability to scale out – Linked Servers are single-threaded.

Read the whole thing. Ben asks and answers the question of whether PolyBase replaces ETL. You’ll want to read his answer. My answer (and I won’t tell you how close it is to his because I want you to read his article) is that PolyBase will only replace a fraction of total ETL and will act as an ETL process in a larger percentage of cases. I can see a pattern where you virtualize the data as external tables and then connect them together locally to insert into local facts and dimensions, for example. But there are too many things you can do with other ETL platforms which make me say this will never be a full replacement.

Comments closed

Copying Cassandra Data to HDFS

Landon Robinson shows how you can use Spark to extract data from Cassandra and move it into HDFS:

Cassandra is a great open-source solution for accessing data at web scale, thanks in no small part to its low-latency performance. And if you’re a power user of Cassandra, there’s a high probability you’ll want to analyze the data it contains to create reports, apply machine learning, or just do some good old fashioned digging.

However, Cassandra can prove difficult to use as an analytical warehouse, especially if you’re using it to serve data in production around the clock. But one approach you can take is quite simple: copy the data to Hadoop (HDFS).

Read on to learn how.

Comments closed

Arrays in Azure Data Factory

Rayis Imayev takes us through arrays in Azure Data Factory:

Currently, there are 3 data types supported in ADF variables: String, Boolean, and Array. The first two are pretty easy to use: Boolean for logical binary results and String for everything else, including the numbers (no wonder there are so many conversion functions in Azure Data Factory that we can use).

I’ve also blogged about using Variables in Azure Data Factory:
– Setting Variables in Azure Data Factory Pipelines
– Append Variable activity in Azure Data Factory: Story of combining things together  
– System Variables in Azure Data Factory: Your Everyday Toolbox 
– Azure Data Factory: Extracting array first element

Click through for arrays and follow up with those other posts from there.

Comments closed

Building an ARM Template for Azure Data Factory

Andy Leondard takes the first steps to building an Azure Data Factory pipeline using Azure Resource Manager Templates:

Azure Resource Manager, or ARM, “allows you to provision your applications using a declarative template.” So says the Azure Quickstart Templates page. ARM templates are JSON and allow administrators to import and export Azure resources using varying management patterns. I really like ARM templates for implementing infrastructure as code in Azure. In this post I show a very simple example of how to use ARM templates to export and then import a basic ADF (Azure Data Factory) pipeline.

The sample code doesn’t do that much by itself, but it does open up a new world of automation.

Comments closed

Minimal Logging with FastLoadContext

Paul White takes us through another way to perform minimally logged bulk loads with SQL Server:

This post provides new information about the preconditions for minimally logged bulk load when using INSERT...SELECT into indexed tables.

The internal facility that enables these cases is called FastLoadContext. It can be activated from SQL Server 2008 to 2014 inclusive using documented trace flag 610. From SQL Server 2016 onward, FastLoadContext is enabled by default; the trace flag is not required.

Without FastLoadContext, the only index inserts that can be minimally logged are those into an empty clustered index without secondary indexes, as covered in part two of this series. The minimal logging conditions for unindexed heap tables were covered in part one.

Click thorugh for a highly informative article.

Comments closed

Populating a Data Vault Model with Azure Data Factory

Rayis Imayev gives us an example of ELT into a Data Vault model using Azure Data Factory:

To make a full transition from the existing  DW model to an alternative Data Vault I removed all Surrogate Keys and other attributes that are only necessary to support Kimball data warehouse methodology. Also, I needed to add necessary Hash keys to all my Hub, Link and Satellite tables. The target environment for my Data Vault would be SQL Azure database and I decided to use a built-in crc32 function of the Mapping Data Flow to calculate hash keys (HK) of my business data sourcing keys and composite hash keys of satellite tables attributes (HDIFF).

Data Vault is somewhere on my list of things to learn. It’s not at the top of the list, but that’s not a slight against it.

Comments closed