Big Data has finally arrived and where it hasn’t, it will soon since there’s no way to unplug the inevitable and mighty avalanche. Just like mobility and BYOD proved compelling forces, Big Data too has become crucial to a business’ strategy and plans. No brand can sideline Big Data, not with new gen consumers producing petabytes of social data for businesses to dive into every day.
So what’s the next step? Making sense of the Big Data shambles. But what with? Graph analytics, the new buzzword that’s doing the rounds in boardrooms across the globe for finer predictive modeling of human/consumer behavior.
Understanding graph analysis
You know social and you know graphs. Combine the two and you get a graph with millions of nodes or vertexes and trillions of edges, reflecting every relationship and experience of a user with other social media users or external entities that link with the social platform. External vertexes can include B2C communities, B2B supply chains, and enterprise applications. It’s easy to see that a graph representing even a small handful of users can reach mindboggling proportions, definitely beyond the analytic capabilities of the average human brain.
Which brings us to social graph analysis and social graph models – meant to identify and discover likely behaviors of people based on their social interactions and actions plotted with vertexes and edges. Graph analysis is not a new stream. It has been used in fields such as engineering and data sciences since many years.
However, social graph analysis has only recently gained popularity and at a fast pace. A number of private, public and research organizations are building social graph analysis tools and leveraging them in real situations such as medicine, research, fraud detection, trading, sentiment monitoring, and more.
Obviously, handling gargantuan proportions of data comes with latency issues and resource contention. But there’s the challenge – supporting graph analysis solutions that can scale, minimize latency, and accelerate response time. Ureka, an enterprise-ready big data appliance from YarcData (subsidiary of Cray) promises to achieve this with up to 512 terabytes of memory onboard for in-memory processing. According to the NSA Big Graph Experiment, graph problems can leverage excellent advances in architecture and libraries if they fit into memory.
Here, there and everywhere
Big data graph analysis holds the center stage everywhere. Google has Pregel, a graph processing system that is part of PageRank. GraphLab (open source project) is a graph analysis and machine learning solution that incorporates algorithms from recommender systems. It purports to support graph databases at leading social media organizations such as Facebook and Twitter.
Hadoop’s MapReduce also fails to deliver when it comes to quick analysis and queries on big unstructured data sets. So the company is introducing YARN aka MapReduce 2.0 which will incorporate traditional parallel-processing methods such as MPI to graph processing to newly developed stream-processing engines such as Storm and S4.
Big data graph analysis is growing into a huge market segment and it won’t be long before IT industry giants jump on board to devise best techniques to manage and analyze Big Data with its three Vs – volume, velocity and variety. That graph analysis will soon become a must across businesses has certainly become evident.So, eager to learn why your business MUST take notice of Mobility ? Or want to decide which app is a right fit for your business? Download your choice !