Skip to main content

What is Shuffling in Spark

Shuffling in Spark is a mechanism that Re-Distributes the data across different executors or workers in the clusters. 

Why do we need to Re-Distribute the data? 

 A) Re-Distribution is needed when there is a need of increasing or decreasing the data partitions in the situations below:

  • When the partitions are not sufficient enough to process the data load in the cluster
  • When the partitions are too high in numbers that it creates task scheduling overhead and it becomes the bottleneck in the processing time.
Re-Distribution can also be achieved by executing the shuffling on existing distributed data collection like RDD, DataFrames, etc by using the "Repartition" and "Coalesce" APIs in Spark.

B) During Aggregation and Joins on data collection in Spark, all the data records belonging to aggregation or join should reside in the single partition and when the existing partitioning scheme doesn't satisfy this condition there is a need to re-distributing the data in input collections before performing the aggregations or joins.

Spark Shuffling is an expensive process as it is moving around data among different executors or workers in the cluster. Imagine, if you have 1000s of workers and have a huge volume of data, in this case, the Spark Shuffling can lead to expensive overheads.

Generally, Spark Shuffling involves
  • Disk I/O
  • Network I/O
  • Data Serialization & Deserialization
Some of the operations that might lead to the Shuffling are 

  • join
  • cogroup
  • groupWith
  • join
  • leftOuterJoin
  • rightOuterJoin
  • groupByKey
  • reduceByKey
  • repartition
  • coalesce
Data Engineers should always consider the possibility where they can avoid Shuffling in their process. For Example, they can choose to use reduceByKey instead of groupByKey due to the following reasons.

  • groupByKey shuffles all the data and therefore it is too slow.
  • reduceByKey shuffles only the sub-aggregation results in the partitions.

To summarize, Shuffling is one of the important operations of Spark that re-distributes the data to make the pipelines more efficient. But we are also aware of how expensive Shuffling could be especially when the number of workers in a cluster is too high or the volume of data is huge. So one should avoid the Shuffling in their pipelines. Also at the same time, one should also understand the concept clearly that could help them to design Fault-tolerant, Robust, and Reliable Data Pipelines.

Comments

Post a Comment

Popular posts from this blog

Best Practices for Data Quality in Data Engineering: Tips and Strategies

Introduction: Data engineering is a critical aspect of modern businesses that rely on data-driven decision-making. However, the effectiveness of data engineering depends on the quality of data it produces. Poor data quality can lead to incorrect decisions, wasted resources, and lost opportunities. Therefore, it's important to implement best practices for data quality in data engineering. In this blog post, we will discuss the tips and strategies for ensuring data quality in data engineering. 1. Establish Data Governance: Data governance refers to the process of defining policies, procedures, and standards for data management. By establishing data governance, you can ensure that data is accurate, complete, and consistent across the organization. This can be achieved through the use of data quality rules, data validation, and data cleansing techniques. 2. Define Data Architecture: Data architecture is the blueprint that outlines the structure of data within an organization. By defini...
1 comment
Read more

DataOps: The Future of Data Engineering

In recent years, a new approach to data engineering has emerged, known as DataOps. This approach emphasizes collaboration, automation, and continuous integration and delivery, and is becoming increasingly popular in organizations that rely heavily on data to drive their business operations. In this post, we'll explore the concept of DataOps, and why it is becoming the future of data engineering. What is DataOps? DataOps is an approach to data engineering that draws inspiration from the DevOps movement in software development. Like DevOps, DataOps emphasizes collaboration and communication between different teams and stakeholders, as well as automation and continuous delivery. In the context of data engineering, this means breaking down silos between data engineers, data scientists, business analysts, and other stakeholders, and creating a culture of shared responsibility for data quality, accuracy, and security. One of the key principles of DataOps is the idea of continuous integra...
Post a Comment
Read more

How to use Cloud Function and Cloud Pub Sub to process data in real-time

Cloud Functions is a fully-managed, serverless platform provided by Google Cloud that allows you to execute code in response to events. Cloud Pub/Sub is a messaging service that allows you to send and receive messages between services. You can use Cloud Functions and Cloud Pub/Sub together to build event-driven architectures that can process data in real-time. Here is a high-level overview of how to use Cloud Functions with Cloud Pub/Sub: Create a Cloud Pub/Sub topic: The first step is to create a Cloud Pub/Sub topic that you will use to send and receive messages. You can do this using the Cloud Console, the Cloud Pub/Sub API, or the gcloud command-line tool. Create a Cloud Function: Next, you will need to create a Cloud Function that will be triggered by the Cloud Pub/Sub topic. You can create a Cloud Function using the Cloud Console, the Cloud Functions API, or the gcloud command-line tool. When you create a Cloud Function, you will need to specify the trigger type (in this case, C...
Post a Comment
Read more