The massive world of Big Data
If one strolls around any IT office premises, over every decade (nowadays time span is even lesser, almost every 3-4 years) one would overhear professionals discussing new jargons from the hottest trends in technology. Around 5 -6 years ago, one such word has started ruling IT services is ‘BIG data’ and still has been interpreted by a layman to tech geeks in various ways.
Although services industries started talking about big data solutions widely from 5-6 years, it is believed that the term was in use since the 1990s by John Mashey from Silicon Graphics, whereas credit for coining the term ‘big data’ aligning to its modern definition goes to Roger Mougalas from O’Reilly Media in 2005.
Let’s first understand why everyone going gaga about ‘BIG data’ and what are the real-world problems it is supposed to solve and then we will try to answer what and how aspects of it.
Why is Big Data essential for today’s digital world?
Pre smart-phones era, internet and web world were around for many years, but smart-phones made it mobile with on-the-go usage. Social Media, mobile apps started generating tons of data. At the same time, smart-bands, wearable devices ( IoT, M2M ), have given newer dimensions for data generation. This newly generated data became a new oil to the world. If this data is stored and analyzed, it has the potential to give tremendous insights which could be put to use in numerous ways.
You will be amazed to see the real-world use cases of BIG data. Every industry has a unique use case and is even unique to every client who is implementing the solutions. Ranging from data-driven personalized campaigning (you do see that item you have browsed on some ‘xyz’ site onto Facebook scrolling, ever wondered how?) to predictive maintenance of huge pipes across countries carrying oils, where manual monitoring is practically impossible. To relate this to our day-to-day life, every click, every swipe, every share and every like we casually do on social media is helping today’s industries to take future calculated business decisions. How do you think Netflix predicted the success of ‘House of Cards’ and spent $100 million on the same? Big data analytics is the simple answer.
Talking about all this, the biggest challenge in the past was traditional methods used to store, curate and analyze data, which had limitations to process this data generated from newer sources and which were huge in volumes generated from heterogeneous sources and was being generated really fast(To give you an idea, roughly 2.5 quintillion data is generated per day as on today – Refer infographic released by Domo called “Data Never Sleeps 5.0.” ), Which given rise to term BIG data and related solutions.
Understanding Big Data: Experts’ viewpoint
BIG data literally means massive data (loosely > 1TB) but that’s not the only aspect of it. Distributed data or even complex datasets which could not be analyzed through traditional methods can be categorized into ‘Big data’ and hence Big data theoretical definition makes a lot of sense with this background:
“Gartner (2012) defines, Big data is high-volume, high-velocity and/or high-variety information assets that demand cost-effective, innovative forms of information processing that enable enhanced insight, decision making, and process automation.”
Generic data possessing characteristics of big data are 3Vs namely Variety, Velocity, and Volume
But due to the changing nature of data in today’s world and to gain most insights of it, 3 more Vs are added to the definition of BIG DATA, namely Variability, Veracity and Value.
The diagram below illustrates each V in detail:
Diagram: 6 V’s of Big Data
This 6Vs help understanding the characteristics of “BIG Data” but let’s also understand types of data in BIG Data processing.
“Variety” of above characteristics caters to different types of data can be processed through big data tools and technologies. Let’s drill down a bit for understanding what those are:
- Structured ex. Mainframes, traditional databases like Teradata, Netezza, Oracle, etc.
- Unstructured ex. Tweets, Facebook posts, emails, etc.
- Semi/Multi structured or Hybrid ex. E-commerce, demographic, weather data, etc.
As the technology is advancing, the variety of data is available and its storage, processing, and analysis are made possible by big data. Traditional data processing techniques were able to process only structured data.
Now, that we understand what big data and limitations of old traditional techniques are of handling such data, we could safely say, we need new technology to handle this data and gain insights out of it. Before going further, do you know, what were the traditional data management techniques?
Traditional Techniques of Data Processing are:
- RDBMS (Relational Database Management System)
- Data warehousing and DataMart
On a high level, RDBMS catered to OLTP needs and data warehousing/DataMart facilitated OLAP needs. But both the systems work with structured data.
I hope. now one can answer, ‘what is big data?’ conceptually and theoretically both.
So, it’s time that we understand how it is being done in actual implementations.
only storing of “big data” will not help the organizations, what’s important is to turn data into insights and business value and to do so, following are the key infrastructure elements:
- Data collection
- Data storage
- Data analysis and
- Data visualization/output
All major big data processing framework offerings are based on these building blocks.
And in an alignment of the above building blocks, following are the top 5 big data processing frameworks that are currently being used in the market:
1. Apache Hadoop: Hadoop software library is a framework that allows for the distributed processing of large data sets across clusters of computers using simple programming models. First up is the all-time classic, and one of the top frameworks in use today. So prevalent is it, that it has almost become synonymous with Big Data.
2 Apache Spark: unified analytics engine for large-scale data processing.
Apache Spark and Hadoop are often contrasted as an "either/or" choice, but that isn't really the case.
Above two frameworks are popular but apart from that following 3 are available and are comparable frameworks:
3. Apache Storm: free and open source distributed real-time computation system. You can also take up Apache Storm training to learn more about Apache Storm.
4. Apache Flink: streaming dataflow engine, aiming to provide facilities for distributed computation over streams of data. Treating batch processes as a special case of streaming data, Flink is effectively both batch and real-time processing framework, but one which clearly puts streaming first.
5. Apache Samza : distributed Stream processing framework.
Frameworks help processing data through building blocks and generate required insights. The framework is supported by the whopping number of tools providing the required functionality.
Big Data processing frameworks and technology landscape
Big data tools and technology landscape can be better understood with layered big data architecture. Give a good read to a great article by Navdeep singh Gill on XENONSTACK for understanding the layered architecture of big data.
By taking inspiration from layered architecture, different available tools in the market are mapped to layers to understand big data technology landscape in depth. Note that, layered architecture fits very well with infrastructure elements/building blocks discussed in the above section.
Few of the tools are briefed below for further understanding:
1. Data Collection / Ingestion Layer
- Cassandra: is a free and open-source, distributed, wide column store, NoSQL database management system designed to handle large amounts of data across many commodity servers, providing high availability with no single point of failure.
- Kafka: is used for building real-time data pipelines and streaming apps. Event streaming platform
- Flume: log collector in Hadoop
- HBase: columnar database in Hadoop
2. Processing Layer
- Pig: scripting language in the Hadoop framework
- MapReduce: processing language in Hadoop
3. Data Query Layer
- Impala: Cloudera Impala: modern, open source, distributed SQL query engine for Apache Hadoop. (often compared with hive)
- Hive: Data Warehouse software for data Query and analysis
- Presto: Presto is a high performance, distributed SQL query engine for big data. Its architecture allows users to query a variety of data sources such as Hadoop, AWS S3, Alluxio, MySQL, Cassandra, Apache Kafka, and MongoDB
4. Analytical Engine
- TensorFlow: n source machine learning library for research and production.
5. Data storage Layer
- Ignite: open-source distributed database, caching and processing platform designed to store and compute on large volumes of data across a cluster of nodes
- Phoenix: hortonworks: Apache Phoenix is an open source, massively parallel, relational database engine supporting OLTP for Hadoop using Apache HBase as its backing store
- PolyBase: s a new feature in SQL Server 2016. It is used to query relational and non-relational databases (NoSQL). You can use PolyBase to query tables and files in Hadoop or in Azure Blob Storage. You can also import or export data to/from Hadoop.
- Sqoop: ETL tool
- Big data in EXCEL: Few people like to process big datasets with current excel capabilities and it's known as Big Data in Excel
6. Data Visualization Layer
- Microsoft HDInsight: Azure HDInsight is a Hadoop service offering hosted in Azure that enables clusters of managed Hadoop instances. Azure HDInsight deploys and provisions Apache Hadoop clusters in the cloud, providing a software framework designed to manage, analyze, and report on big data with high reliability and availability. Hadoop administration training will give you all the technical understanding required to manage a Hadoop cluster, either in a development or a production environment.
Best Practices in Big Data
Every organization, industry, business, may it be small or big wants to get benefit out of “big data” but it's essential to understand that it can prove of maximum potential only if organization adhere to best practices before adapting big data:
Answering 5 basic questions help clients know the need for adapting Big Data for organization.
- Try to answer why Big Data is required for the organization. What problem would it help solve?
- Ask the right questions.
- Foster collaboration between business and technology teams.
- Analyze only what is required to use.
- Start small and grow incrementally.
Big Data industry use-cases
We talked about all the things in the Big Data world except real use cases of big data. In the starting, we did discuss few but let me give you insights into the real world and interesting big data use cases and for a few, it’s no longer a secret ☺. In fact, it’s penetrating to the extent you name the industry and plenty of use cases can be told. Let’s begin.
1. Streaming Platforms
As I had given an example of ‘House of Cards’ at the start of the article, it’s not a secret that Netflix uses Big Data analytics. Netflix spent $100mn on 26 episodes of ‘House of Cards’ as they knew the show would appeal to viewers of original British House of Cards and built in director David Fincher and actor Kevin Spacey. Netflix typically collects behavioral data, and it then uses this data to create a better experience for the user.
But Netflix uses Big Data for more than that, they monitor and analyze traffic details for various devices, spot problem areas and adjust network infrastructure to prepare for future demand. (later is action out of Big Data analytics, how big data analysis is put to use). They also try to get insights into types of content viewers to prefer and help them make informed decisions.
Apart from Netflix, Spotify is also a known great use case.
2. Advertising and Media / Campaigning /Entertainment
For decades marketers were forced to launch campaigns while blindly relying on gut instinct and hoping for the best. That all changed with digitization and big data world. Nowadays, data-driven campaigns and marketing is on the rise and to be successful in this landscape, a modern marketing campaign must integrate a range of intelligent approaches to identify customers, segment, measure results, analyze data and build upon feedback in real time. All needs to be done in real time, along with the customer’s profile and history, based on his purchasing patterns and other relevant information and Big Data solutions are the perfect fit.
Event-driven marketing is also could be achieved through big data, which is another way of successful marketing in today’s world. That basically indicates, keeping track of events customer are directly and indirectly involved with and campaign exactly when a customer would need it rather than random campaigns. For. Ex if you have searched for a product on Amazon/Flipkart, you would see related advertisements on other social media apps you casually browse through. Bang on, you would end up purchasing it as you anyway needed options best to choose from.
3. Healthcare Industry
Healthcare is one of the classic use case industries for Big Data applications. The industry generates a huge amount of data.
Patients medical history, past records, treatments given, available and latest medicines, Medicinal latest available research the list of raw data is endless.
All this data can help give insights and Big Data can contribute to the industry in the following ways:
- Diagnosis time could be reduced, and exact requirement treatment could be started immediately. Most of the illnesses could be treated if a diagnosis is perfect and treatment can be started in time. This can be achieved through evidence-based past medical data available for similar treatments to doctor treating the illness, patients’ available history and feeding symptoms real-time into the system.
- Government Health department can monitor if a bunch of people from geography reporting of similar symptoms, predictive measures could be taken in nearby locations to avoid outbreak as a cause for such illness could be the same.
The list is long, above were few representative examples.
4. Security
Due to social media outbreak, today, personal information is at stake. Almost everything is digital, and majority personal information is available in the public domain and hence privacy and security are major concerns with the rise in social media. Following are few such applications for big data.
- Cyber Crimes are common nowadays and big data can help to detect, predicting crimes.
- Threat analysis and detection could be done with big data.
5. Travel and Tourism
Flight booking sites, IRCTC track the clicks and hits along with IP address, login information, and other details and as per demand can do dynamic pricing for the flights/ trains. Big Data helps in dynamic pricing and mind you it’s real time. Am sure each one of us has experienced this. Now you know who is doing it :D
Telecommunications, Public sector, Education, Social Media and gaming, Energy and utility every industry have implemented are implementing several of these Big Data use cases day in and day out. If you look around am sure you would find them on the rise.
Big Data is helping everyone industries, consumers, clients to make informed decisions, whatever it may be and hence wherever there is such a need, Big Data can come handy.
Challenges faced by Big Data in the real world for adaptation.
Although the world is going gaga about big data, there are still a few challenges to implement and adopt Big Data and hence service industries are still striving towards resolving those challenges to implement best Big Data solution without flaws.
An October 2016 report from Gartner found that organizations were getting stuck at the pilot stage of their big data initiatives. "Only 15 percent of businesses reported deploying their big data project to production, effectively unchanged from last year (14 per cent)," the firm said.
Let’s discuss a few of them to understand what are they?
1. Understanding Big Data and answering Why for the organization one is working with.
As I started the article saying there are many versions of Big Data and understanding real use cases for organization decision makers are working with is still a challenge. Everyone wants to ride on a wave but not knowing the right path is still a struggle. As every organization is unique thus its utmost important to answer ‘why big data’ for each organization. This remains a major challenge for decision makers to adapt to big data.
2. Understanding Data sources for the organization
In today’s world, there are hundreds and thousands of ways information is being generated and being aware of all these sources and ingest all of them into big data platforms to get accurate insight is essential. Identifying sources is a challenge to address.
It's no surprise, then, that the IDG report found, "Managing unstructured data is growing as a challenge – rising from 31 per cent in 2015 to 45 per cent in 2016."
Different tools and technologies are on the rise to address this challenge.
3. Shortage if Big Data Talent and retaining them
Big Data is changing technology and there are a whopping number of tools in the Big Data technology landscape. It is demanded out of Big Data professionals to excel in those current tools and keep up self to ever-changing needs. This gets difficult for employees and employers to create and retain talent within the organization.
The solution to this would be constant upskilling, re-skilling and cross-skilling and increasing budget of organization for retaining talent and help them train.
4. The Veracity V
This V is a challenge as this V means inconsistent, incomplete data processing. To gain insights through big data model, the biggest step is to predict and fill missing information.
This is a tricky part as filling missing information can lead to decreasing accuracy of insights/ analytics etc.
To address this concern, there is a bunch of tools. Data curation is an important step in big data and should have a proper model. But also, to keep in mind that Big Data is never 100% accurate and one must deal with it.
5. Security
This aspect is given low priority during the design and build phases of Big Data implementations and security loopholes can cost an organization and hence it’s essential to put security first while designing and developing Big Data solutions. Also, equally important to act responsibly for implementations for regulatory requirements like GDPR.
6. Gaining Valuable Insights
Machine learning data models go through multiple iterations to conclude on insights as they also face issues like missing data and hence the accuracy. To increase accuracy, lots of re-processing is required, which has its own lifecycle. Increasing accuracy of insights is a challenge and which relates to missing data piece. Which most likely can be addressed by addressing missing data challenge.
This can also be caused due to unavailability of information from all data sources. Incomplete information would lead to incomplete insights which may not benefit to required potential.
Addressing these discussed challenges would help to gain valuable insights through available solutions.
With Big Data, the opportunities are endless. Once understood, the world is yours!!!!
Also, now that you understand BIG DATA, it's worth understanding the next steps:
Gary King, who is a professor at Harvard says “Big data is not about the data. It is about the analytics”
You can also take up Big Data and Hadoop training to enhance your skills furthermore.
Did the article help you to understand today’s massive world of big data and getting a sneak peek into it Do let us know through the comment section below?
