What keeps you relevant in any industry? Your skill-set. And when it comes to computer science, the ever-changing technologies and softwares require updating of skill-sets more frequently than in any other industry. Hadoop is one such emerging or rather emergent software which is in huge demand nowadays.
While being able to learn its operation and usage is good enough for getting your hands on some projects, Hadoop interview questions could get a little tricky if you miss out on the basic stuff. A minor survey about Hadoop interview questions revealed that the trend of questions leans more towards the application part rather than the theoretical part of the software.
Here are 5 Hadoop interview questions and answers that one should not miss to make your interview a cakewalk.
1. What is Apache Hadoop?
• To store and analyse large sets of unstructured data, Apache Hadoop is used. It is like the backbone of all Big Data applications. Since it is an open source tool, it can handle huge amounts of data that is written in Java.
Apache Hadoop has three components:
a. HDFS- Short for Hadoop Distributed File System, it is primary data storage system that is used in Hadoop. It is a data management layer along with Yarn Hadoop. HDFS is a Java based file system.
b. MapReduce- Used for distributed processing of huge sets of data, MapReduce is a software framework in computer clusters.
c. Yarn- Apache Yarn Hadoop is a resource management layer that splits up functionalities for resource management. This architectural centre enables handling of data stored in a single platform by the use of multiple data processing engines like interactive SQL, data science, real time streaming and batch processing.
2. What do you understand by Data Locality in Hadoop?
• Developed as a mechanism to tackle a fundamental issue in the Hadoop system, Data Locality prevents cross stitching of network traffic due to the huge volume of data. Data Locality moves Map codes or Map tasks (computation) closer to the data.
All data in Hadoop is stored inside the HDFS (Hadoop Distributed File System). HDFS splits the data spreading it across the network. Data is used to perform a task when you submit it. In case the data is too far from the task, network issue could arise. To overcome this issue, “Data Locality” helps move the map tasks closer to the data.
3. What does Safemode mean in the context of Hadoop?
• The maintenance of NameNode in Apache Hadoop is Safemode- no modifications allowed in the file system during this time. This makes the HDFS cluster a read-only, not allowing it to delete or even replicate Data Blocks.
Data Blocks are stored in NameNode that contains details such as location and replica. This meta-data is stored in memory for its faster retrieval. NameNodes maintains and manages the slave nodes and assigns tasks to them. It is this NameNode, the modification to which is not allowed in Safemode.
How to check the status of Safemode?
Hadoopdfs admin –safemode get
How does one enter Safemode?
bin/hadoopdfs admin –safemode enter
How does one come out of Safemode?
Hadoop dfsadmin –safe mode leave
4. In which all modes can Hadoop be run?
There are 3 modes in which Hadoop can run. These are:
• Local or Standalone mode- Hadoop in the default mode runs in the local mode or standalone mode. Input and output operations in this mode are done using the local file system. It is used in the debugging process but does not support HDFS. Custom configuration for configuration files is not required in Standalone mode.
• Pseudo Distributed Mode (Single Node Cluster)- Unlike in local mode, in this mode custom configuration is required. But similar to Standalone mode, Pseudo Mode runs on a single node in a pseudo distributed mode. Also, both the master and slave node are the same in Pseudo mode.
• Fully Distributed Mode (Multi Node Cluster)- In Fully Distributed Mode, data is used and is distributed across various nodes on a Hadoop cluster. Here, the master and slave nodes are separate. Multi cluster nodes are formed in this mode because all daemons perform in distinct nodes.
5. What is rack awareness? What are its advantages?
• Hadoop components are rack aware. The understanding of how various data nodes are distributed across the racks of a Hadoop cluster is known as rack awareness. It is required in Hadoop for various reasons. It improves data availability and reliability, improves the cluster’s performance, and the network bandwidth. It prevents the data from getting lost in case of rack failure. However, the chance of node failure is higher than that of rack failure. Rack awareness also helps us keep the bulk data in rack when possible. Another plus is that rack awareness minimises writing time and increases reading speed. It does so by placing write/read requests to replicas on nearby or the same rack.
Looking for Hadoop Jobs in top cities? Click at the links below: