Nokia, as the global leader in mobile services, decided to turn its attention to the data age of mobilitythe practice of collecting and analyzing big data to transform the user experience. More recently, the industry-wide wave of leveraging terabytes of unstructured data particularly inspired this company to explore suitable hardware platforms for building a big data analytics infrastructure.
In Nokia, terabytes of unstructured data from mobiles, services, or log files are streamed continuously, but capturing this large-scale data within a RDBMS framework was cost prohibitive. Also, in such a scenario, the data types would be highly limited. We knew wed break the bank trying to capture all this unstructured data in a structured environment, said Amy OConnor, Senior Director of Analytics at Nokia.
Nokia’s multinational business outfit typically requires mining volumes of reliable, market insights or information on collective behaviors of groupsfrom its enterprise data to first migrate and then analyze the data in a structured environment with sophisticated analytical tools. These kinds of complex, data-analytics environmental requirements could not be solved with their existing model of distributed silos; so they began a search for a better solution.
Nokia’s search for an enterprise solution
A company connecting more than 1.3 billion people was not about to settle for a mediocre solution of wrestling out fragmented intelligence from individual silos catering to limited needs. The company gradually came to the realization that if they wanted to derive enterprise value from the collective strength of all the data contained in these silos, they required a single, integrated data-capturing and -storing platform to conduct high-end data analytics. They needed a solution that could address endless volumes of unstructured data along with structured data, in a centralized data-analysis location which could yield very high quality business intelligence by cross referencing multi-functional enterprise data.
Nokia’s big data processing frequently requires predictive traffic maps or layered location information; so the technology solution for their big data analytics center had to be the best.
O’Connor quipped Nokia differentiates itself based on the data we have.
The ultimate solution: Cloudera Hadoop with Teradata
The technical staff at Nokia was already familiar with Hadoop, and several of them were experimenting with Clouderas Distribution of Apache Hadoop (CDH). While exploring Clouderas offering, this company realized that Hadoop represented a cost-friendly and dependable, data-capturing and -storage platform for multi-structured data. Additionally, this platform facilitated high-speed data retrieval and processing of petabytes of data without any loss of data quality. CDH has simplified the process of deploying and implementing this system. With Hadoop, the average cost, per terabyte of storage is, at least 10 times cheaper than a relational data warehouse. This distribution of Hadoop also supported the reformatting of unstructured data into a relational schema before processing.
Although Hadoop perfectly fitted Nokias data analytics infrastructural requirements, the in-house expertise on this platform was nothing to be desired. So, Nokia partnered with Cloudera to strengthen their existing engineering and technical support teams to build a scalable solution that fits their brand image.
So how did Teradata contribute to the big picture?
Teradata provided an enterprise data warehouse (EDW), which, along with Oracle and MySQL data marts, formed the core of this technology ecosystem. Nokia evenly distributed its terabytes of structured data assets on Teradata EDW, and volumes of multi-structured data on the Hadoop Distributed File System (HDFS). The centralized Hadoop cluster in UK data center contains 0.5 PB of data.
Nokia’s enterprise data continuously moves from data warehouses and marts to a multi-tenant Hadoop cluster, enabling the company’s 60,000+ employees to tap into this priceless data.