Data Warehouse Interview Question and Answer

1. Define Data Warehousing in simple words.

Ans. This is one of the commonly asked data warehouse interview questions which you can answer by saying that –

Data warehousing can be called a repository of data, which helps management teams in driving apt business decisions.

It is a process that involves data collection and data management which helps provide significant insights to businesses. Being the core of Business Intelligence data warehouse analyst is one of the most sought after careers in 2021. Today, the data warehouse is an essential practice for almost every industry, including verticals like healthcare IT, automation, retail, logistics and government agencies. 

2. Are databases and data warehousing the same thing?

Ans. The database is a way of storing information in an organized format, and it is represented in the form of a table, columns, and rows. Nowadays, companies use dedicated database management software (DBMS) to store crucial data.

While data warehousing also helps in storing data, it is used to store a large chunk of data, and it allows users to use data for complex queries. For this, users take the help of Online Analytical Processing (OLAP).

Though both databases and data warehouses are relational data systems, they serve different purposes. Below are some key differences:

3. How is OLTP different from OLAP?

Ans. OLTP stands for Online Transactional Processing, which deals with current data and is characterized by short write transactions. The main objective of OLTP is to record all the current updates, insertions, and deletions, and thus, it is less time-consuming and easy to maintain. Also, OLTP acts as a source of data for OLAP.

The focus of OLAP is to store historical data that has been processed by OLTP. OLAP helps in data analysis and supports in reaching out to meaningful interpretations. Some of the noticeable differences between OLTP and OLAP are:

4. What are some benefits of cloud-based data warehouses?

Ans. Below are the top reasons why companies prefer to use cloud-based data warehouses over traditionally used on-site warehouses –

Flexible and dynamic infrastructure – You will have a range of infrastructure available at your perusal and you can explore further to find out the optimal infrastructure for your business.

Secure data – Cloud data is much more secure and low on cost. Cloud encryption services like multi-factor authentication make data transportation even more secure.

Increased data usage – Cloud data warehousing provides flexibility and agility. With a cloud data warehouse, it is easier to control the cluster size, CPU, and RAM so that they fit the requirements of the unique projects.

Business capabilities – Data warehousing on the cloud offers better business capabilities like scalability, availability, disaster recovery, extensibility, accessibility, flexibility, execution capacity, security, etc.

Optimized for data analytics – The cloud data warehouse is optimized for data analytics because it uses Massively Parallel Processing (MPP) and columnar storage, which are known for offering better performance and helps in executing complex queries

Lower cost of ownership – With a very low cost of capital, flexibility, and infrastructure capabilities, the risks of running a data warehouse project are reduced and the chances of success are increased, along with the recused costs of ownership. 

5. Name essential approaches to data warehouse design.

Ans. There are two data warehouse design approaches, Kimball and Inmon.

Inmon approach or top-down was proposed by Mr. Bill Inmon, the Father of data warehousing. In this approach, first, it is recommended to prepare a data warehouse, and then Data Marts are created. Through this strategy, the data warehouse becomes the central point of the Corporate Information Factory (CIF), which acts as a logical framework for BI.

Kimball approach, also known as a bottom-up approach, suggests creating Data Mart first and later integrating it to a more massive data warehouse to complete a data warehouse. This integration of Data Mart is known as a data warehouse bus (BUS) architecture.

6. What are the advantages and disadvantages of the Inmon approach?

Ans. Below are some advantages and disadvantages of top-down or Inmon design:

7. What are the advantages and disadvantages of the Kimball approach?

Ans. Below are some advantages and disadvantages of bottom-up or Kimball design:

8. Which are the different types of data warehousing?

Ans. There are three types of data warehousing:

1. Enterprise Data Warehouse

It merges organizational data from its different functional areas in a centralized manner. It helps with data extracting and transforming and offers a detailed overview of any particular object in the data model.

1. Operational Data Store

It gives an option to produce the data from the database instantly and supports business operations by integrating contrast data from multiple sources.

1. Data Mart

Data Mart stores data from a particular functional area, and it comprises a subset of data that is saved in the data warehouse. It helps the analyst in swiftly analyzing the data by shrinking the volume of a large chunk of data.

9. Name 3 types of Data Mart.

Ans. Below are the 3 different types of Data Marts:

1. Dependent – It sources organizational data from a single data warehouse and helps in developing more Data Marts.
2. Independent – Here, no data is dependent on a central or enterprise data warehouse, and data can be used separately for conducting an independent analysis.
3. Hybrid – It helps in ad hoc integration and is used when a data warehouse comprises inputs from different sources. 

10. What is data warehouse architecture?

Ans. Conceptualized with a relational database management system (RDBMS), data warehouse architecture serves as a central repository for informational data. Here, the central repository includes several key components that make the environment operative, compliant, and accessible to operational systems.

11. What is the three-tier architecture of a data warehouse?

Ans. Below is the three-tier data warehouse architecture:

Image Source – Digital Vidya

1. Bottom Tier

It represents the data warehouse database server, which is also known as the relational database system. It uses backend tools and utilities that are used to feed data and perform functions like – Extract, Clean, Load, and Refresh.

1. Middle Tier

It represents the OLAP Server, which is a form of the extended relational database management system. It is known to implement multidimensional data and operations.

1. Top Tier

It factors the front-end client layer and holds query, analysis, and data mining tools.

12. What are the different stages of data warehouse decision support evolution?

Ans. Below are the 5 stages involved in data warehouse decision support evolution:

1. Report
2. Analyze
3. Predict
4. Operationalize
5. Active warehousing 

13. Name the components of data warehousing.

Ans. Below are the 5 components of data warehousing:

1. Data Warehouse Database
2. Sourcing, Acquisition, Clean-up, and Transformation Tools (ETL)
3. Metadata
4. Query Tools
5. Data warehouse Bus Architecture

14. Name some of the popular data warehouse tools.

Ans. Below is the list of popular query tools:

15. What do you know about Amazon Redshift’s architecture?

Ans. Amazon Redshift, based on PostgreSQL, is the most popular cloud service offered by Amazon Web Services. This tool is popularly used for handling Petabyte-scale data. Its unique features help the analyst to query data in seconds. With almost negligible cost, Redshift is easy to set up and maintain.

Redshift can be integrated with other BI and analytical tools and works with Extract, Transform, and Load (ETL) tools.

Below are some features of Redshift:

1. Columnar storage and MPP processing
2. Compression (column-level operation)
3. Management and Security
4. Data Types
5. Updates and Upserts

16. State something about real-time data warehousing.

Ans. Real-time data warehousing is a concept, which reflects the real-time state of the warehouse by capturing the data as soon as it occurs. It has low latency data, which is fast, scalable, and simple to use.

17. What are the benefits of real-time data warehousing?

Ans. Below are some benefits of using real-time data warehousing:

• Eases decision making
• Resolves the problem of ideal data load
• Ensures quick recovery and permits more rapid interventions
• Eliminates batch window
• Easy to optimize by running transformations in the database

18. What should you avoid when planning to construct a real-time data warehouse?

Ans. One must avoid mistakes like:

• Not focusing on data integrity when constructing real-time data
• Overlooking traditional OLTP systems
• Not initiating business process changes in real-time data warehousing 

19. What do you mean by SCD?

Ans. SCD stands for a slowly changing dimension, which is used to store and manage historical data. It is among the most critical tasks that support tracking dimension record history. 

20. Which are the three types of SCD?

Ans. Below are the three types of slowly changing dimension:

• 1st Layer – SCD 1 – Overwriting current record with the new record
• 2nd Layer – SCD 2 – Creating another dimension record to an existing customer dimension table
• 3rd Layer – SCD 3 – Creating a current value field to include new data 

21. Define Schema in data warehousing.

Ans.

22. State the difference between Star and Snowflake schema.

Ans. Below is the list of differences between star schema and snowflake schema:

23. Define a Galaxy schema.

Ans. Galaxy schema, also known as Fact Constellation Schema, contains two fact tables along with dimensional tables. In other words, it can be called a combination of stars.

24. What are the types of fact tables?

Ans. In the dimensional model, the fact table is the primary table, which contains facts and foreign keys to the dimension table. It is used for measurement in the business process. The fact table has three different types:

25. What are the types of dimension tables?

Ans. Joined via a foreign key, a dimension table includes the dimension of facts. It is also known as denormalized tables that offer descriptive characteristics of facts. Below are the types of dimension tables:

• Conformed dimensions
• Outrigger dimensions
• Shrunken rollup dimensions
• Dimension-to-dimension table joins
• Junk dimensions

26. Give the steps to start and shut down the database.

Ans. Below are the steps to start a database:

• Start an instance
• Mount the database
• Open the database

Below are the steps to shut down a database:

• Close the database
• Dismount the database
• Shutdown the instance 

27. Define the surrogate key.

Ans. Surrogate key functions as a substitute for the natural primary key.

28. What do you mean by virtual data warehousing?

Ans. It is a collective view of the finished data, and it does not include historical data. The main objective of the virtual data warehouse is to help in making analytical decisions making and translating raw data into a more presentable format. Along with this, it also offers a semantic map.

29. Define XMLA.

Ans. XMLA or XML for Analysis is the Simple Object Access Protocol, which is used as a standard for obtaining data in OLAP. 

30. Differentiate between View and Materialized View.

Ans. Below table highlights the difference between view and materialized view:

31. When do you use bteqexport?

Ans. Whenever the total number of rows is less than half a million, bteqexport is used. 

32. When do you use fastexport?

Ans. “fastexport” is used when the total number of rows is more than half a million. 

33. Name the primary functions of dimensions.

Ans. The primary functions of the dimensions are:

• Filtering
• Grouping
• Labeling 

34. As a data warehouse manager, what were your key job responsibilities in the previous company?

Ans. Some of my prime responsibilities are:

• Work on creating data warehouse process models
• Verify the integrity of warehouse data and ensure consistent changes
• Implement data extraction procedures
• Maintain data standards
• Handle data related troubleshooting
• Use different computer language and methods to perform data analysis
• Implement metadata processes
• Review data designs, codes, and test plans
• Use database management system software like Apache, MongoDB, Oracle to smoothly perform data warehousing functions 

35. Which data warehousing skills did you master?

Ans. In the data warehousing interview, you can talk about your critical technical skills. You can say – some of my strengths are:

• Enterprise system management software
• Apache Avro
• Human resource management software HRMS
• Data mining software like Rapid-I RapidMiner, SAP NetWeaver Business Warehouse
• CRM software
• Data Analysis
• MS Office

36. What is Metadata?

Ans. Metadata consists of information that characterizes data, describes the content, quality, conditions, history, availability, and other characteristics of the data. Metadata allows a person to locate and understand data, including information required to determine what data sets for a particular geographic location, information necessary to determine if a data set is appropriate for specific purposes, information required to retrieve or get an already identified set of data and the information required to process and use it.

37. What are the benefits of metadata?

Ans. Primary benefits of Metadata include – 

Facilitates search and analysis – Metadata helps to find data more easily and allows data analysis from the source itself, favoring self-documentation, transformation, and reporting, among other functions.

Improves data governance – Managing metadata in a standardized environment ensures good data governance, contributing to the success of the program.

Integration aid: In hybrid integration, metadata is key. Using a centralized metadata repository for shared use between IT and business users facilitates governance, as well as the application of best practices. Metadata is very useful in hybrid structures as it improves data management in an integrated way.

Facilitates standardization – Metadata eliminates errors and improves the quality of metadata throughout its life cycle, along with a complete vision of said cycle, from start to end.

Improves reporting – Metadata management improves reporting, allowing them to be delivered safely and reliably. This is due to the ease of intervention that makes the processes of higher quality.

Interoperability – Metadata facilitates interoperability since metadata standards have been defined and there are shared protocols for the exchange of this information. Protocols such as Z39.50 or CSW have helped in simultaneous searches for data in distributed systems.

38. How does a data cube help?

Ans. Data cubes help us represent data in multiple dimensions. Data cubes are defined by dimensions and facts and are created from a subset of attributes in the database.

39. What is a load manager?

Ans. A load manager performs the necessary operations to extract and load the process. The size and complexity of the load manager vary between specific solutions from the data warehouse to the data warehouse.

Load manager helps to pull data from the source system and quickly load the extracted data into the temporary data warehouse. It also performs simple transformations in a structure similar to that of the data warehouse. It is supported by underlying DBMS and allows client programs to generate SQL to be executed at a server.  

40. What do you mean by data extraction?

Ans. Data extraction is the process of extracting data captured within semi-structured and unstructured sources such as emails, PDF documents, PDF forms, text files, social media, barcodes, and images. Data extraction is done with the help of enterprise-grade data extraction tools, making incoming business data from unstructured or semi-structured sources usable for data analysis and reporting.

Structured formats can be processed directly in most business intelligence tools after some debugging. However, an ideal PDF Scraping Tool should also support common unstructured formats, including DOC, DOCX, PDF, TXT, and RTF, allowing businesses to make use of all the information they receive.

41. What is the difference between Data Extraction Vs Data Mining?

Ans. The data extraction process deals with extracting important information from sources such as emails, PDF documents, forms, text files, social media, barcodes, and images with the help of content extraction tools. Data mining, on the other hand, is a process used to look for patterns, anomalies, and correlations in the data. Data mining tools allow users to analyze data from multiple perspectives to identify hidden patterns in large data sets.

42. What Does Junk Dimension Mean?

Ans. Junk dimensions are the single dimensions, used to store the tiny dimensions known as junk attributes. The junk attributes are a set of text attributes and flags that are transmitted into a different sub-domain known as the junk dimension.

We can consider the example of car colors and car bodies. As we can see that these attitudes are limited in number and, if created as single dimensions, the dimensions would be limited to a single attribute. To eliminate these small dimensions, we create a single “junk” dimension that cross joins all possible attributes into a single dimension to be used in the fact table.

43. What is data replication?

Ans. Data replication is the process of copying and storing business data at multiple locations. The replication process can be single or continuous, depending on the organization’s requirements.

The primary purpose of data replication is to improve the availability and accessibility of the data and the robustness and consistency of the system. Data replication works by copying data from one location to another, for example, between two local hosts in the same location or in different locations.   

44. What are the benefits of database replication?

Ans. The main benefits of database replication are –

Increased reliability – By replicating the database across multiple servers, you ensure that the data will be available even in case of a serious hardware failure. The distributed database management system is able to route affected users to another of the available nodes.

Performance improvement – Since the data is distributed on different servers, the multiple accesses do not saturate the servers. This is especially important for applications that can have thousands or hundreds of thousands of simultaneous requests, leading to increased performance.

Improved data security – In a traditional transactional system, all updates to a database are saved on the same disk. The security of data is then in the hands of the backup strategy implemented on that server. With database replication, data security is increased since updates are written on multiple servers, such as disks, several power supplies, CPUs, etc.

Data analysis support – Typically, data-driven companies duplicate data from numerous sources in their data warehouses, such as data warehouses or data lakes. This makes it easier for the analytics team dispersed in multiple locations to undertake shared business intelligence projects.

45. What is dimensional modeling?

Ans. Dimensional modeling is a way of bringing data to the way these will be converted into useful information for business users. The ultimate goal is that they can intuitively and quickly find the information they need.

The application of the dimensional model takes place in the logical design phase, which allows the translation of the resulting schema from the conceptual design to the logical plane.  This technique is widely accepted and is often chosen as the preferred one to represent analytical data because it simultaneously meets the following requirements:

• Arrange and structure the data in a way that is understandable to the business user
• Generates high performance in searches from the reporting layer