Press "Enter" to skip to content

Category: Spark

Technology Choices for Streaming Pipelines

Published 2020-05-08 by Kevin Feasel

The Hadoop in Real World team takes us through different tools available when working on streaming pipelines:

Businesses want to get insights as quickly as possible and do not want to wait for a day, like before, to bring up a report to understand what happened till yesterday. They require a more proactive approach that can help to act immediately when something significant happens and also to prevent the system from any faults/downtime before it occurs. Imagine you are buying some product from an e-retailer and you have gone till the point to make payment and something happened that caused the payment not to go through successfully. At that very moment, you are having a second thought about whether to buy the product now or later. Suppose, if the business is getting a report of this occurrence next day, it would not be of much use for them as the customer would have already bought it from somewhere or decided against it. This is where real-time events and insights come in. If it were a real-time report, the team would have called up the customer and made the purchase by offering some discounts, which in turn would have changed the mind of the customer.

Click through for a high-level discussion of these tools.

Comments closed

Security Practices for Azure Databricks

Published 2020-05-07 by Kevin Feasel

Abhinav Garg and Anna Shrestinian walk us through good security practices when using Azure Databricks:

Azure Databricks is a Unified Data Analytics Platform that is a part of the Microsoft Azure Cloud. Built upon the foundations of Delta Lake, MLflow, Koalas and Apache SparkTM, Azure Databricks is a first party PaaS on Microsoft Azure cloud that provides one-click setup, native integrations with other Azure cloud services, interactive workspace, and enterprise-grade security to power Data & AI use cases for small to large global customers. The platform enables true collaboration between different data personas in any enterprise, like Data Engineers, Data Scientists, Business Analysts and SecOps / Cloud Engineering.

In this article, we will share a list of cloud security features and capabilities that an enterprise data team could utilize to bake their Azure Databricks environment as per their governance policy.

Much of this is fairly straightforward, but it is nice to have it all in one place.

Comments closed

Dynamic File Pruning on Delta Lake

Published 2020-05-06 by Kevin Feasel

Ali Afroozeh, et al, take us through Dynamic File Pruning in Databricks Runtime 6.1:

In addition to eliminating data at partition granularity, Delta Lake on Databricks dynamically skips unnecessary files when possible. This can be achieved because Delta Lake automatically collects metadata about data files managed by Delta Lake and so, data can be skipped without data file access. Prior to Dynamic File Pruning, file pruning only took place when queries contained a literal value in the predicate but now this works for both literal filters as well as join filters. This means that Dynamic File Pruning now allows star schema queries to take advantage of data skipping at file granularity.

There are some interesting performance results here. I’d also be curious to see how robust the results are as queries get more complicated

Comments closed

Reading Query Plans in Spark

Published 2020-05-01 by Kevin Feasel

Daniel Ciocirlan has a primer on query plans in Apache Spark:

Let’s go over some examples of query plans and how to read them. Let’s go back to the one we’ve just shown:

 == Physical Plan == *(1) Project [(id#0L * 5) AS id#2L]

+- *(1) Range (1, 1000000, step=1, splits=6)

We read this plan backwards, bottom to top:

Spark does have some UI components which make this a bit easier, but you’ll probably end up in a situation where you need to read it in this format.

Comments closed

Developing for Databricks with VS Code

Published 2020-04-29 by Kevin Feasel

Gerhard Brueckl tells us what comes after notebooks for users with development backgrounds:

For those users Databricks has developed Databricks Connect (Azure docs) which allows you to work with your local IDE of choice (Jupyter, PyCharm, RStudio, IntelliJ, Eclipse or Visual Studio Code) but execute the code on a Databricks cluster. This is awesome and provides a lot of advantages compared to the standard notebook UI. The two most important ones are probably the proper integration into source control / git and the ability to extend your IDE with tools like automatic formatters, linters, custom syntax highlighting, …

While Databricks Connect solves the problem of local execution and debugging, there was still a gap when it came to pushing your local changes back to Databricks to be executed as part of a regular ETL or ML pipeline. So far you had to either “deploy” your changes by manually uploading them via the Databricks UI again or write a script that uploads it via the REST API (Azure docs).

Gerhard has a nice extension for Visual Studio Code which helps with this. I’m also a huge fan of the DatabricksPS module, so I’ll happily plug that here.

Comments closed

Handling Bad Records with Apache Spark

Published 2020-04-06 by Kevin Feasel

Divyansh Jain shows three techniques for handling invalid input data with Apache Spark:

Most of the time writing ETL jobs becomes very expensive when it comes to handling corrupt records. And in such cases, ETL pipelines need a good solution to handle corrupted records. Because, larger the ETL pipeline is, the more complex it becomes to handle such bad records in between. Corrupt data includes:

– Missing information
– Incomplete information
– Schema mismatch
– Differing formats or data types

Since ETL pipelines are built to be automated, production-oriented solutions must ensure pipelines behave as expected. This means that data engineers must both expect and systematically handle corrupt records.

This is the seedy underbelly of semi-structured data: you don’t have control over the data as it comes in, so you have to control the data coming out.

Comments closed

Tips for Moving from Pandas to Koalas

Published 2020-04-03 by Kevin Feasel

Haejoon Lee, et al, walk us through migrating existing code written for Pandas to use the Koalas library:

In particular, two types of users benefit the most from Koalas:

– pandas users who want to scale out using PySpark and potentially migrate codebase to PySpark. Koalas is scalable and makes learning PySpark much easier
– Spark users who want to leverage Koalas to become more productive. Koalas offers pandas-like functions so that users don’t have to build these functions themselves in PySpark

This blog post will not only demonstrate how easy it is to convert code written in pandas to Koalas, but also discuss the best practices of using Koalas; when you use Koalas as a drop-in replacement of pandas, how you can use PySpark to work around when the pandas APIs are not available in Koalas, and when you apply Koalas-specific APIs to improve productivity, etc. The example notebook in this blog can be found here.

Read on to learn more.

Comments closed

Data Exfiltration Protection when Using Azure Databricks

Published 2020-03-31 by Kevin Feasel

Bhavin Kukadia, et al, explain how to prevent users from taking data from your Databricks cluster without authorization:

Solving for data exfiltration can become an unmanageable problem if the PaaS service requires you to store your data with them or it processes the data in the service provider’s network. But with Azure Databricks, our customers get to keep all data in their Azure subscription and process it in their own managed private virtual network(s), all while preserving the PaaS nature of the fastest growing Data & AI service on Azure. We’ve come up with a secure deployment architecture for the platform while working with some of our most security-conscious customers, and it’s time that we share it out broadly.

Click through for the architectural pattern.

Comments closed

Schema Management for Spark Applications

Published 2020-03-27 by Kevin Feasel

Walaa Eldin Moustafa takes us through some of the things that LinkedIn has learned about schema management with Apache Spark:

At LinkedIn, the Hive Metastore is the source of truth catalog for all Hadoop data. The Hive Metastore is managed by Dali. Dali is a data access and processing platform that is integrated to compute engines and ETL pipelines at LinkedIn to ensure consistency and uniformity in the access and storage of data. Dali utilizes the Hive Metastore to store data formats, data locations, partition information, and table information. Among other features, Dali also manages the definition of SQL views, as well as storing and accessing those definitions from the Hive Metastore.

Read on for a good explanation of the how as well as the why.

Comments closed