Press "Enter" to skip to content

Category: Hadoop

Stream Reactor Update

Published 2018-01-23 by Kevin Feasel

Andrew Stevenson announces Stream Reactor 1.0.0 for Kafka Connect 1.0:

Stream Reactor is an Apache License, Version 2.0 open source collection of components built on top of Kafka and provides Kafka Connect compatible connectors to move data between Kafka and popular data stores. Stream Reactor provides source connectors to publish data into Kafka and sink connectorsto bring data from Kafka into other systems. The connectors support KCQL (Kafka Connect Query Language), an open source component of Lenses SQL Enginethat provides an elegant and simple SQL like syntax for selecting fields and routing from sources or topics to Kafka or the target system (topic to target entity mapping, field selection, auto creation, auto evolution, error policies).

We hope you find Stream Reactor useful, and want to give it a try! Stream Reactor has over 25 connectors available, tested and documented, supporting both Kafka 0.11 and Kafka 1.0 and you can give it a go by downloading Lenses Development Environment or find the jars on GitHub, or even build the code locally and help us improve and add even more connectors.

Read on for more details, as well as a link to the GitHub repo.

Comments closed

Unit Testing Spark Streaming DStreams

Published 2018-01-22 by Kevin Feasel

Anuj Saxena shows how to create unit tests for DStreams in Spark Streaming:

The method ‘ testOperation ‘ takes the output of the operation performed on the ‘inputPair’ and check whether it is equal to the ‘outputPair’ and just like this, we can test our business logic.

This short snippet lets you test your business logic without forcing you to create even a Spark session. You can mock the whole streaming environment and test your business logic easily.

This was a simple example of unary operations on DStreams. Similarly, we can test binary operations and window operations on DStreams.

Click through for an example with code.

Comments closed

Kafka Topic Reuse

Published 2018-01-19 by Kevin Feasel

Martin Kleppmann walks through the trade-offs of reusing Apache Kafka topics for different event types:

The common wisdom (according to several conversations I’ve had, and according to a mailing list thread) seems to be: put all events of the same type in the same topic, and use different topics for different event types. That line of thinking is reminiscent of relational databases, where a table is a collection of records with the same type (i.e. the same set of columns), so we have an analogy between a relational table and a Kafka topic.

The Confluent Avro Schema Registry has traditionally reinforced this pattern, because it encourages you to use the same Avro schema for all messages in a topic. That schema can be evolved while maintaining compatibility (e.g. by adding optional fields), but ultimately all messages have been expected to conform to a certain record type. We’ll revisit this later in the post, after we’ve covered some more background.

For some types of streaming data, such as logged activity events, it makes sense to require that all messages in the same topic conform to the same schema. However, some people are using Kafka for more database-like purposes, such as event sourcing, or exchanging data between microservices. In this context, I believe it’s less important to define a topic as a grouping of messages with the same schema. Much more important is the fact that Kafka maintains ordering of messages within a topic-partition.

Read the whole thing.

Comments closed

Set Operations In Spark

Published 2018-01-18 by Kevin Feasel

Fisseha Berhane compares SparkSQL, DataFrames, and classic RDDs when performing certain set-based operations:

In this fourth part, we will see set operators in Spark the RDD way, the DataFrame way and the SparkSQL way.
Also, check out my other recent blog posts on Spark on Analyzing the Bible and the Quran using Spark and Spark DataFrames: Exploring Chicago Crimes.

The data and the notebooks can be downloaded from my GitHub repository.
The three types of set operators in RDD, DataFrame and SQL approach are shown below.

This is where SparkSQL (and SQL in general) shines, although the DataFrame approach is also compact.

Comments closed

Setting Up SparklyR In Azure

Published 2018-01-17 by Kevin Feasel

David Smith shows how you can spin up a Spark cluster in Azure and install SparklyR on top of it:

The SparklyR package from RStudio provides a high-level interface to Spark from R. This means you can create R objects that point to data frames stored in the Spark cluster and apply some familiar R paradigms (like dplyr) to the data, all the while leveraging Spark’s distributed architecture without having to worry about memory limitations in R. You can also access the distributed machine-learning algorithms included in Spark directly from R functions.

If you don’t happen to have a cluster of Spark-enabled machines set up in a nearby well-ventilated closet, you can easily set one up in your favorite cloud service. For Azure, one option is to launch a Spark cluster in HDInsight, which also includes the extensions of Microsoft ML Server. While this service recently had a significant price reduction, it’s still more expensive than running a “vanilla” Spark-and-R cluster. If you’d like to take the vanilla route, a new guide details how to set up Spark cluster on Azure for use with SparklyR.

Read on for more details.

Comments closed

Apache NiFi 1.5 Updates

Published 2018-01-17 by Kevin Feasel

Tim Spann shows off some nice additions to Apache NiFi:

Another cool processor that I will talk about in greater detail in future articles is the much-requested Spark Processor. The ExecuteSparkInteractive processor with its Livy Controller Service gives you a much better alternative to my hacky REST integration to calling Apache Spark batch and machine learning jobs.

There are a number of enhancements, new processors, and upgrades I’m excited about, but the main reason I am writing today is because of a new feature that allows for having an Agile SDLC with Apache NiFi. This is now enabled by Apache NiFi Registry. It’s as simple as a quick git clone or download and then, you’ll use Apache Maven to install Apache NiFi Registry and start it. This process will become even easier with future Ambari integration for a CLI-free install.

To integrate the Registry with Apache NiFi, you need to add a Registry Client. It’s very simple to add the default local one — see below.

There are several new features in the latest release.

Comments closed

How Meltdown And Spectre Have Affected Spark Performance

Published 2018-01-17 by Kevin Feasel

Chris Stevens, et al, show how DAtabricks customers have fared in a post-Meltdown+Spectre world:

On AWS, we have observed a small performance degradation up to 5% since January 4th. On i3-series instance types, where we cache data on the local NVMe SSDs (Databricks Cache), we have observed a degradation up to 5%. On r3-series instance types, in which the benchmark jobs read data exclusively from remote storage (S3), we have observed a smaller increase of up to 3%. The greater percentage slowdown for the i3 instance type is explained by the larger number of syscalls performed when reading from the local SSD cache.

The chart below shows before and after January 3rd in AWS for a r3-series (memory optimized) and i3-series (storage optimized) based cluster.  Both tests fixed to the same runtime version and cluster size. The data represents the average of the full benchmark’s runtime per day, for a total of 7 days prior to January 3 (before is in blue) and 7 days after January 3 (after is in red). We exclude January 3rd to prevent partial results.  As mentioned, the i3-series has the Databricks Cache enabled on the local SSDs, resulting in roughly half of the total execution time (faster) compared to the r3-series results.

Overall, they’re seeing a degredation of 2-5%.  Click through for some more information on how they collected their metrics.

Comments closed

Comparing Data Lake Job Runs

Published 2018-01-17 by Kevin Feasel

Yanan Cai shows how to compare stats on different executions of a job:

Troubleshooting issues in recurring job is a time-consuming task. It starts with searching through the Job Browser to find instances of a recurring job and identifying both baseline and anomalous performance. This is followed by multi-way comparisons between job instances to figure out what has been changed in the query, data or environment. This is followed by analysis to discover which changes may have performance impact. While this is happening production workloads continue to under-perform or go down.

Azure Data Lake Tools for Visual Studio now makes it easy to spot anomalies and quickly trace the key characteristics across recurring job instances allowing for an efficient debugging experience. The Pipeline Browser automatically groups recurring jobs to simplify discovery of all runs. The Related Job View collects data about inputs, outputs and execution across multiple runs into a single visualization.

Read on for more.

Comments closed

Active Directory And ElasticMapReduce

Published 2018-01-11 by Kevin Feasel

Bruno Faria shows how to use AWS’s CloudFormation to extend Active Directory into an AWS ElasticMapReduce cluster and run jobs via Kerberos:

In this example, you build a solution that allows Active Directory users to seamlessly access Amazon EMR clusters and run big data jobs. Here’s what you need before setting up this solution:

  • An AWS account
  • An Amazon EC2 key pair
  • A possible limit increase for your account (Note: Usually a limit increase will not be necessary. See the AWS Service Limits documentation if you encounter a limit error while building the solution.)

To make it easier for you to get started, I created AWS CloudFormation templates that automatically configure and deploy the solution for you. The following steps and resources are involved in setting up the solution:

  1. Create and configure an Amazon Virtual Private Cloud (Amazon VPC).
  2. Launch an Amazon EC2 Windows instance (Active Directory domain controller).
  3. Create an Amazon EMR security configuration for Kerberos and cross-realm trust.
  4. Launch an Amazon EMR cluster with Kerberos enabled and a cross-realm trust configuration.

You can use the AWS CloudFormation templates to complete each step individually, or you can deploy the entire solution through a single step.

Read the whole thing.

Comments closed

Spark And NVMe

Published 2018-01-11 by Kevin Feasel

Alicja Luszczak, et al, introduce NVMe caching in the Databricks distribution of Spark:

A particularly important and widespread use case is caching the results of scan operations. This allows the users to eliminate the low throughput associated with reading remote data. For this reason, many users who intend to run the same or similar workload repeatedly decide to invest extra development time into manually optimizing their application, by instructing Spark exactly what files to cache and when to do it, and thus “explicit caching.”

For all its utility, Spark cache also has a number of shortcomings. First, when the data is cached in the main memory, it takes up space that could be better used for other purposes during query execution, for example, for shuffles or hash tables. Second, when the data is cached on the disk, it has to be deserialized when read — a process that is too slow to adequately utilize the high read bandwidths commonly offered by the NVMe SSDs. As a result, occasionally Spark applications actually find their performance regressing when turning on Spark caching.

Third, having to plan ahead and explicitly declare which data should be cached is challenging for the users who want to interactively explore the data or build reports. While Spark cache gives data engineers all the knobs to tune, data scientist often find it difficult to reason about the cache, especially in a multi-tenant setting, where engineers still require the results to be returned as quickly as possible in order to keep the iteration time short.

Read on for more details, as well as performance comparisons.

Comments closed