Book Image

Big Data Forensics: Learning Hadoop Investigations

By : Joe Sremack
Book Image

Big Data Forensics: Learning Hadoop Investigations

By: Joe Sremack

Overview of this book

Big Data forensics is an important type of digital investigation that involves the identification, collection, and analysis of large-scale Big Data systems. Hadoop is one of the most popular Big Data solutions, and forensically investigating a Hadoop cluster requires specialized tools and techniques. With the explosion of Big Data, forensic investigators need to be prepared to analyze the petabytes of data stored in Hadoop clusters. Understanding Hadoop’s operational structure and performing forensic analysis with court-accepted tools and best practices will help you conduct a successful investigation. Discover how to perform a complete forensic investigation of large-scale Hadoop clusters using the same tools and techniques employed by forensic experts. This book begins by taking you through the process of forensic investigation and the pitfalls to avoid. It will walk you through Hadoop's internals and architecture, and you will discover what types of information Hadoop stores and how to access that data. You will learn to identify Big Data evidence using techniques to survey a live system and interview witnesses. After setting up your own Hadoop system, you will collect evidence using techniques such as forensic imaging and application-based extractions. You will analyze Hadoop evidence using advanced tools and techniques to uncover events and statistical information. Finally, data visualization and evidence presentation techniques are covered to help you properly communicate your findings to any audience.
Table of Contents (10 chapters)
9
Index

Collecting HBase evidence


HBase differs from Hive in a number of ways. First, HBase is not a relational database. Unlike Hive, HBase does not support SQL-like queries, because SQL is a language for relational databases. Second, HBase does not have a metastore database. Instead, HBase is a nonrelational database based on Google's BigTable that works with HDFS for data storage and access. Third, HBase data is distributed to various nodes in regions, or to data blocks that store column-oriented chunks of related data. It is far easier to collect HBase evidence through HBase rather than collecting from each node due to the distributed nature of the data.

Given the complexity of carving out data from HFiles, collecting HBase evidence through the HBase interface has an advantage over a filesystem collection. HFiles are distributed file structures that need to be collected from each node. Once collected, HFiles must be carved in order to extract the column-oriented data and metadata and then convert...