Press "Enter" to skip to content

Category: ETL / ELT

Serverless Lambda Architecture

Published 2017-09-12 by Kevin Feasel

Laith Al-Saadoon shows off a new Amazon Web Services product, AWS Glue, which allows you to build a data processing system on the Lambda architecture without directly provisioning any EC2 instances:

With the launch of AWS Glue, AWS provides a portfolio of services to architect a Big Data platform without managing any servers or clusters. AWS Glue is a fully managed extract, transform, and load (ETL) service that makes it easy for customers to prepare and load their data for analytics. You can create and run an ETL job with a few clicks in the AWS Management Console. You simply point AWS Glue to your data stored on AWS, and AWS Glue discovers your data and stores the associated metadata (for example, the table definition and schema) in the AWS Glue Data Catalog. After it’s cataloged, your data is immediately searchable, queryable, and available for ETL.

AWS Glue generates the code to execute your data transformations and data loading processes. Furthermore, AWS Glue provides a managed Spark execution environment to run ETL jobs against a data lake in Amazon S3. In short, you can now run a Lambda Architecture in AWS in a completely 100% serverless fashion!

“Serverless” applications allow you to build and run applications without thinking about servers. What this means is that you can now stream data in real-time, process huge volumes of data in S3, and run SQL queries and visualizations against that data without managing server provisioning, installation, patching, or capacity scaling. This frees up your users to spend more time interpreting the data and deriving business value for your organization.

Laith has a working demo of the process available as well.

Comments closed

Streaming ETL Using CDC And Event Hub

Published 2017-07-12 by Kevin Feasel

Rolf Tesmer combines Change Data Capture and Event Hubs to build a streaming ETL solution:

The solution picks up the SQL data changes from the CDC Change Tracking system tables, creates JSON messages from the change rows, and then posts the message to an Azure Event Hub.  Once landed in the Event Hub an Azure Stream Analytics (ASA) Job distributes the changes into the multiple outputs.

What I found pretty cool was that I could transmit SQL delta changes from source to target in as little as 5 seconds end to end!

There are a bunch of steps, but the end result is worth it.

Comments closed

Real-Time Streaming ETL With Kafka Streams

Published 2017-06-27 by Kevin Feasel

Yeva Byzek has a tutorial using Kafka and Kafka Streams to perform real-time ETL:

Let’s consider an application that does some real-time stateful stream processing with the Kafka Streams API. We’ll run through a specific example of the end-to-end reference architecture and show you how to:

  • Run a Kafka source connector to read data from another system (a SQLite3 database), then modify the data in-flight using Single Message Transforms (SMTs) before writing it to the Kafka cluster

  • Process and enrich the data from a Java application using the Kafka Streams API (e.g. count and sum)

  • Run a Kafka sink connector to write data from the Kafka cluster to another system (AWS S3)

Read the whole thing.

Comments closed

Re-Shaping Data Flows

Published 2017-06-21 by Kevin Feasel

Maneesh Varshney explains some methods to trim the fat out of analytical data flows:

Big data comes in a variety of shapes. The Extract-Transform-Load (ETL) workflows are more or less stripe-shaped (left panel in the figure above) and produce an output of a similar size to the input. Reporting workflows are funnel-shaped (middle panel in the figure above) and progressively reduce the data size by filtering and aggregating.

However, a wide class of problems in analytics, relevance, and graph processing have a rather curious shape of widening in the middle before slimming down (right panel in the figure above). It gets worse before it gets better.

In this article, we take a deeper dive into this exploding middle shape: understanding why it happens, why it’s a problem, and what can we do about it. We share our experiences of real-life workflows from a spectrum of fields, including Analytics (A/B experimentation), Relevance (user-item feature scoring), and Graph (second degree network/friends-of-friends).

The examples relate directly to Hadoop, but are applicable in other data platforms as well.

Comments closed

Change Detection With Hashes

Published 2017-06-14 by Kevin Feasel

Nigel Meakins shows how to use HashBytes to roll your own change detection:

So this all sounds very promising as a way of tracking changes to our Data Warehouse data, for purposes such as extracting deltas, inserts and updates to Type I and II dimensions and so forth. It doesn’t have any show-stopping overhead for the hashing operations for the sizes of data typically encountered and storage isn’t going to be an issue. It is native to T-SQL so we can rerun our hash value generation in the engine where our data resides rather than having to push through SSIS or some other tool to generate this for us. Algorithms are universal and as such will give us the same values wherever used for the same bytes of input. Let’s go back to the basic idea for a minute and consider how we implement this.

This is particularly useful in cases where you have metadata columns you don’t much care about (e.g., last modified time).  I do recommend using CONCAT or CONCAT_WS (if you’re on SQL Server 2017) to do string concatenation, though; it’d remove the need for util.CastAsNVarchar and possibly more.

Comments closed

Mapping File Shenanigans With The Import/Export Wizard

Published 2017-06-08 by Kevin Feasel

Angela Henry ran into problems copying a boatload of data from a mainframe-hosted DB2 server and has lived to tell the tale:

This post talks about the issue I ran into with SSIS Mapping Files.

We currently run DB2 on an IBM iSeries AS400 for our ERP system.  I was tasked with copying data from the AS400 to a SQL Server database for some consultants to use.  The C-Suite didn’t want to give the consultants access to our AS400, so this was the work around that was put forth and accepted (and no, no one asked me before I was “voluntold” for the task).  Since this would essentially be a “one-time” thing, I chose to use the Import Export Wizard, but I would save the package just in case they wanted this process repeated.

Sounds like it was a painful experience, but it does have a happy ending.

Comments closed

JSON Data Sources In SSIS

Published 2017-06-07 by Kevin Feasel

Chris Koester shows how to read JSON data sources in SQL Server Integration Services:

Once the Script Component has been defined as a source, the output columns can be defined. For this post, the same USGS Earthquake data that was used in the “Download JSON data with PowerShell” post will serve as an example. Be careful to choose the correct data types here. This can be tedious because you have to choose the correct data types in the C# code as well, and ensure that they correspond with the SSIS types. It’s helpful to bookmark a SSIS data type translation table for reference.

It does involve creating a script component, but aside from the tedium that Chris mentions, it’s not too bad.

1 Comment

Collecting Rows Written From SSISDB

Published 2017-05-24 by Kevin Feasel

Andy Leonard has a script showing how to get the number of records written to an OLEDB destination, as reflected in the SSIS Catalog:

I prefer to passively gather performance metrics – even if it’s a little convoluted and more work (for me). I don’t often need these metrics immediately available, so I execute queries similar to the query below and store the performance metrics in a table. I can (and do) build dashboards to track SSIS performance (perhaps I should blog about them…) using this passively-collected performance metrics.

As with Lookup Transformation messages, OLE DB Destinations in Data Flow Tasks record specific messages using the same format every time. Because of this (hard-coded) consistency, you and I can passively collect the number of rows written while executing packages in the SSIS Catalog using the (default) Basic logging level. We can use the following Transact-SQL query to collect this execution metadata post-execution:

Click through for the script.

Comments closed

The Continued Importance Of ETL

Published 2017-05-08 by Kevin Feasel

Andy Leonard explains that good old ETL remains vital to an organization:

A Problem

As Jen points out earlier in her Analytics Market Commoditization and Consolidation post (you should read it all – it’s awesome – like all of Jen’s posts!) many analytics solution providers share the “Same look, same marketing story, same saves time and allows users [to] avoid evil IT.”

I can hear some of you thinking, “Are you telling us analytics doesn’t work, Andy?” Goodness no. I’m telling you hype and sales strategy work in the analytics market as well as anywhere. When asked why a solution may not perform to expectations, the #1 response is “your data is not clean.”

Data engineering (think ETL specifically designed for analytics and “big data”) is the backbone behind data science.  To Andy’s point, the data engineer’s job is to get clean, context-heavy data in front of a data scientist, the same way a “classical” Business Intelligence specialist works with analysts.

Comments closed

Using Talend To Build Shape Files

Published 2017-05-03 by Kevin Feasel

Paul Hernandez has a demo where he uses Talend’s product to convert latitude and longitude pairs to a shape file:

Input data

Customers coordinates: a flat file containing x,y coordinates for every customer.

Municipalities in Austria: a shape file with multi-polygons defining the municipalities areas in Austria: source

Goal

The goal was to “look-up” the coordinates in the shape file in order to get the municipality code GKZ which in german stand for “Gemeindekennzahl”.

Check out the demo.

Comments closed