Oracle Big Data Series - Part 1
Relational databases have been the backbone for decades, on which Enterprises have been taking business decisions and basing their strategies. For some that data set's been growing leaps and bounds with their business, and global footprint.Careful examination reveals that we are still talking about strucutred data thus far, while there's been a revolutionary change in the unstructured, semi-structured world of Information management.
To counter this we have seen significant changes in the way traditional relational databases have aligned their services, offering and features bringing transactional processing, Data Warehousing specific functionalities, Clustering approaches to handle the increasing loads and a lot to offer for BI initiatives. Careful examination reveals that we are still talking about strucutred data thus far, while there's been a revolutionary change in the unstructured, semi-structured world of Information management. Data is coming from weblogs, social media, emails, smart meters, sensors, videos, images, machine logs in addition to the increase in volumes of traditional relational databases.
Interesting to note that volumes have gone up by multi-fold encompassing those data sources, the expectations on the cost of ownership, building solutions, and the speed of analytics is inversely proportional:
High Volumes -> Lower Cost of ownership
High Volumes and Variety of Data Sources -> Low Cost and faster solution implementations
High Volumes, frequency and # of data sources -> Higher the speed of analytics expectations (this was quite opposite prior to current times of Big Data)
High Velocity of Data -> Reduced Latency of Analysis/Analytics
The next question can be broken into 2 broad categories:
Q1.1 -> As a business do we understand what data sources are critical, relevant and has the most business value that can be un-earthed?
Q1.2 -> What toolsets do i require/invest that optimizes cost and yet provides me a comprehensive solution to achieve my Enterprise goals from Analytics?
The answer to those questions really boils down to defining the right use-cases as a starting point e.g. as a Automobile Manufacturing organization one would be keen to achieve following:
1. Predict faults during manufacturing to reduce the downstream impacts in assembly, and actionize the resolutions at the point of fault occurences - by trending and analyzing the sensor/shop floor logs
2. Analyzing customer service logs, complaints during vehicle services - identifying patterns of common problems in vehicle components by make/model and co-relating them with Manufacturing log analysis
3. Gauging customer sentiments online - Leveraging Web logs, Twitter, Facebook, Blogs etc to bring out sentiments, dis-satisfaction with vehicle, product, services that can be handled before it spreads among the customer network
4. Post sale analyzing the sensor data that's recevied from the vehicle in operations to predict and warn of faults likely to occur - therby reducing service costs, customer satisfaction and vehicle design improvement during manufacture
Are those 4 the only use-cases that Automobile Manufacturer may think of? Certainly not, however this can be start with high impact on overall revenues and customer satisfaction for the manufacturer if those can be converted into a real solution. One needs a comprehensive solution and portfolio of products that can help achive the end goals here. The key buckets for such a solution must have:
1. Acquire - Diverse set of data types
3. Analyze - Provide Insights, Visualization and Dig out hidden relationships
Oracle's Big Data Appliance, and Oracle Big Data Connectors is one such solution and portfolio of products bundled together to provide an integrated solution. The beauty of Oracle Big Data Appliance comes with fact that the process of Acquire, Organize and Analyze work quite closely with their proven Oracle RDBMS technology and thus gives Enterprises flexibility to cover SQL and NoSQL (Key-Value store databases for unstructured, semi-structured data) within the horizon of their analytical needs.
Next blog in the series will have a closer look at Oracle's Big Data Appliance architecture, and key components built for providing a comprehensive Big Data solution.