Big Data
and the 3 V’s:
Volume, Variety
and Velocity
The three V’s are the defining properties of big data. It is critical to understand what these elements mean. The main point of the V-based characterization of big data is to highlight its most serious challenges: the capturing, cleaning, curating, integrating, storing, processing, indexing, searching, sharing, transferring, mining, analyzing, and visualizing of large volumes of fast-moving, highly complex data. The good news is that there are many big data solutions to help out, including the MapR M7 Enterprise Database Edition for Hadoop, which received the highest ranking among big data deployments, according to a recent Forrester report1.
The first V stands for volume.
Volume describes the absolute magnitude of the data being analyzed. Terabyte is becoming a relatively small amount, as petabyte becomes the metric being used to quantify data sets in industries across the board. It is easy to revel in the astronomical numbers around data in this new world, a world measured in zettabytes – 1,180,591,620,717,411,303,424 bytes. Large numbers like these and the amount of processing power needed to digest this amount of information make it easy to capture the attention of early adopters.
The second V stands for variety.
Variety in big data refers to the differing types of data. This is where the discussion turns to Social Media. Buried in the monotony of social data, nuggets of advertising gold may be found. The problem with this is the digging process. The information advertisers and B2C businesses want, is buried in mountains of
non-structured data.
The information one can mine from social media is unlike traditional structured information and as a result, does not fit well into a standard database, where you can run tried and true analytical tools. The information is found in the metadata of photos and videos. It is found in statuses about forgotten anniversaries and
in the well-meant birthday wishes of individuals.
Footnotes:
Even if you limit your data tracking to only text-based information, what one gets is varied. For
example, if one wants to track what people think of ‘orange’ on Twitter, it’s not as simple as just tracking the word orange. There’s ‘orange’ the fruit, the color, the French tech company, the city in California, the county in California and that is just the first page of results from a Google search.
Variety includes structured, database ready information and unstructured data, including different forms of media and seemingly simple, text-based information that may require context to be analyzed properly.
The final V is Velocity.
Velocity is the sheer speed of data. Statistics such as every minute, users watch over 138,000 hours of video on YouTube. Every minute, 27,778 new blog posts go live on Tumblr. Every minute, 100,000 Tweets are shared. Every minute, 208,000 pictures are posted on Facebook. Sometimes, these authors will glaze over more industrial statistics that are equally amazing, such as every flight a Virgin Atlantic Boeing 787 takes collects 500 gigs of data.
This data is analyzed in real time, allowing preparations for repairs to occur before the plane ever lands at its destination. The speed of incoming data is imperative and the life and usefulness of this data is short. To take advantage of it, the data must be analyzed in real time, requiring a huge amount of computing power.
These aspects of big data would cause almost any adopter to pause. However, to continue in the quest towards knowing what big data is and the different ways one can use it, one must wrap ones head around Hadoop.
2http://hadoop.apache.org/
“The Apache Hadoop software library is a framework that allows for the distributed
processing of large data sets across clusters of computers using simple programming
models. It is designed to scale up from single servers to thousands of machines, each
offering local computation and storage. Rather than rely on hardware to deliver
high-availability, the library itself is designed to detect and handle failures at the
application layer, so delivering a highly-available service on top of a cluster of
computers, each of which may be prone to failures.
2”
Hadoop is an open-source software licensed by the Apache Software Foundation. Part of the features of the software is to provide for scalability within big data applications. Traditional architectures have grids that are for storage only, compute only, and report only. This segregated system makes big data applications inflexible.
Data collection has gotten to the point where the volume, velocity and the variety of incoming data has increased the complexity of processing dramatically. Users with traditional architectures are finding their systems cannot finish processing today’s load of data before tomorrow’s load starts flowing in. With a traditional architecture, it is not possible
to add a new server into the architecture and have it help process the query that has been slowing the system down; the computation would have to start over. It is not possible to add new data to the set and continue the computation where you left it; the
computation would have to start over. Hadoop solves the dilemma by grouping servers together and then having them act as one clustered system. Within that capability, Hadoop allows for the addition of new data into an ongoing query and it allows for scaling of the physical infrastructure without
interrupting current processing. Another key benefit to Hadoop is the systematic process of replication that has been designed into the operation. Each data set is broken up into blocks and then those blocks are replicated on 3 different nodes (servers) in the cluster. That means if any one block goes down, it has 2 backups of replication data.
There are also some weaknesses with Hadoop such that ultimately, the solutions being implemented are a hybrid architecture, to blend the traditional approaches around structured and lasting historical data, and the evolving approaches for bursting and unstructured data that need extremely fast processing and analyzing.
Raw Data
Replicated Data
Being analyzed and stored
An important component to big data, which sometimes goes unnoticed, is the physical data center facility, which is vital to the success of big data architectures. The data
center is the backbone for big data applications. For some small companies there are cloud options available. However, for many companies the security of the cloud becomes an issue.
As a result, many companies need to manage their own data and their own server clusters. The
information big data solutions are trying to collect, crunch and store can be extremely private in nature. The health care industry, for example, has many regulations dealing with the collection and use of data, including the Health Insurance Portability and Accountability Act (HIPAA) which establishes procedures for the exercise of individual health information privacy rights. For some companies, the sheer size of their data makes using cloud based solutions uneconomical.
Companies looking to experiment with clusters start with around 100 servers (around 3-4 racks). Adoption leaders have much bigger clusters; Yahoo for example, was reported by Information Week to have 42,000 servers (around 1,400 racks) in their clusters. The space required to store a cluster like that is massive and can start to double in size very rapidly as the data expands. Keeping the clusters cool, powered and accessible to the internet is no easy feat. For many companies, the different aspects of managing a data center are outside of their core competencies. That is why more and more enterprise companies are choosing to outsource their data center needs, by using colocation providers.
Big data provides businesses a level of business intelligence that can set them apart from their competition and give them valuable insights into their market. Despite this fact, Gartner research estimates that 85% of Fortune 500 companies will be unable to exploit big data for competitive advantage.
About CyrusOne
With over two dozen data centers across the globe, CyrusOne helps many of the world’s largest global businesses – including 9 of the global Fortune 20 companies and over 140 of the Fortune 1000 – and companies of all sizes take advantage of the latest data center technology and realize top operational efficiencies through:
• Flexible, Scalable Solutions – Receive flexible data center solutions that readily scale to match the
needs of your growing business.
• Proven, Innovative Technology – Benefit from the latest data center innovations CyrusOne expert
technicians can put to work for your IT environment.
• Exceptional Service – Enjoy personalized, consultative service through all stages of the relationship
- design, build, installation, management, and reporting.
• CyrusOne National IX – Offers low-cost metro connectivity and city-to-city transport in an ever
growing number of cities across the US.
About the Author
Scott Brueggeman oversees the management of CyrusOne’s global marketing,
product development, inside sales, and corporate communications including branding, demand creation, and public relations. His 20 years of marketing and sales experience includes Fortune 50 firms, as well as smaller high-growth companies. Prior to
CyrusOne, he spent several years with running marketing at a data center hosting and managed services company, as well as Chief Marketing Officer at PEAK6 Investments, an international financial services firm. Prior to that he was VP Marketing for
CareerBuilder, and also held leadership positions at AT&T and PepsiCo. Brueggeman serves on several advisory boards.