Using Kafka & Power BI for a Telco

Location: Canada

Industry: Telecom

Problems

Churn rate management: When our client contacted us, it was facing an increase in the subscriber churn rate.
Gaining visibility into the installed infrastructure: The client had many network boosters installed at various places, and these boosters were connected to many devices. A major concern for the customer was to gather the real-time analysis of the time when the signals via boosters dropped, which in turn affected the signals received by its customers.

The customer also required business intelligence reports about:

  • The time when the signals via devices dropped
  • . The time when the signals got revived again
  • The tenure between the above-mentioned points

While being unable to measure the divergence, not only the client’s productivity was getting affected, but its customer churn was also being impacted.

Basically, with this information, they wanted to measure the switching ratio of their clients, as poor services often lead to end-customers defecting to competing broadband service providers.

The challenge was not only to generate such reports. But in doing so, data was to be picked, filtered, and refined from tons of raw data produced via routers.

Solutions

We, developed a big data solution on Cloudera to process RMON files to provide a real-time dashboard.

  • Designed and implemented a solution to process 300 GB data daily on Cloudera.
  • Used Spark to implement a solution to process RMON data for real-time data analysis.
  • Used Filebeat and Log stash to ingest data from a multitude of sources.
  • Used Kafka to build a real-time data pipeline and store streams of records in a fault-tolerant way in Topics.
  • Used Flume to collect, aggregate, and move large amounts of data from Kafka's topic.
  • Used HDFS to store large data sets reliably.
  • Set up of standalone and distributed Hadoop cluster (CDH4, CDH5).
  • Managed Hadoop cluster and its ecosystem including HDFS, Kafka, HBase, Flume, Spark and Zookeeper using Cloudera Manager.
  • We worked to Troubleshoot Hadoop Job failures, slowed down the running queries, and cluster issues.
  • Commissioning, decommissioning, add/remove a node from cluster/ Cloudera manager or rebalancing a Hadoop cluster.
  • Commissioning, decommissioning, add/remove a node from cluster/ Cloudera manager or rebalancing a Hadoop cluster.

Tech Stack

  • Kafka
  • Filebeat
  • Logstash
  • HDFS
  • HBase
  • Spark
  • Scala
  • Flume
  • SSRS
  • Power BI

Get In Touch With Us

we believe in  cutting  edge solutions  and are committed to your  success

Let's Work Together