ZoneVoice is a leading provider of international voice connectivity and solutions to telecom-related businesses. With a large presence in the African market and vast expertise and know-how, ZoneVoice caters to a wide-range of businesses including mobile operators, national regulatory entities, retail services vendors and system integrators.
ZoneVoice manages its own data centers where they store all of the calls' detailed records (cdrs). Recently, the amount of data has been growing exponentially. Meanwhile, they have struggled to deal with the vigorous work required, so they tended to delete the old records and keep the statistics created from these cdrs. The objective was to build a BI dashboard that can visualize cdrs statistics and assist decision makers to perform advanced analysis.
The customer project scope involved helping them transform their reporting and analytics landscape with implementation of an end-to-end solution including data lake, data warehouse and analytics dashboard where the system should automate data feed and should only access the cdrs statistics from the source system. Moreover, the solution should be extensible for future work related to artificial intelligence and machine learning.
We started our data pipeline with a Lambda function that is operating the data ingestion. The Lambda is connected through a JDBC connection to the data source and is able to query the data with the specific date and pull the data to the S3 bucket. Using AWS cloud formation, we are managing the data governance framework and the access list to the data lake. Once the data arrives to S3 bucket, another AWS Lambda function is triggered to run the glue workflow to crawl the data and process to the proper format that is saved in the data warehouse Redshift and Amazon S3. Amazon Athena provides the ability to run ad-hoc queries on the processed data in S3. QuickSight allowed the data analytics team to build an interactive dashboard to assist business analysts at ZoneVoice draw insights from their data and make data-driven decisions as required.
The automation of our solution is based on an AWS Lambda function daily trigger. Once the data is imported to our S3 data lake, another Lambda function is triggered that runs the glue workflow to crawl the data and process it. Athena and Quicksight are directly connected to the respective data sources which means they are online with them.
Specialists at ZoneVoice were using an on-premises high performance computing (HPC) solution to run analytics on their big data, however the model wasn't able to scale due to the resources volumes. Their data specialists faced a host of issues regarding the run times and data covered. With Zero&One, the company refactored their analytics solution on AWS which led to reducing Infrastructure costs by 35%, due to the HPC becoming redundant. Additionally, the witnessed a 25% decrease in run times which allowed them to view their results earlier for more precise decisions. Factoring in the reduction in both areas they are expecting an ROI within 2 years.