From Legacy to Lightning-Fast: How PeerIslands Transformed Telco’s Platform

In this case study, a leading Telco faced scalability challenges with legacy technology. An optimized MongoDB data processing pipeline, leveraging Confluent Kafka, Kotlin, and Spring Boot, delivered impressive results. The solution scaled seamlessly, achieving faster processing, and ensuring 99.999% uptime.

Impact

100X Faster

End to end processing of < 1sec compared to 2 min target

40k/sec

Raw events ingested into time series collection. Aggregate views recomputed

~ 1 Petabyte

Solution scales to handle ~ 1 Petabyte of data

99.999%

99.999% uptime enabled by 5 node replica sets across 3 data centres

~46ms

The above results were demonstrated with cross-region latency of ~46ms

Context

  • With roll out of 5G, Leading Telco expects to handle an exponentially larger number of devices on its Thing space platform (from 50M devices to 100M devices in two years).

  • The current monolithic platform built on legacy technology stack – Tibco EMS, SQL Server (4 servers each with 6 TB RAM) is maxed out supporting 50MM devices generating 12k msgs / sec.

  • Leading Telco has to re-architect to handle business’ growth objectives of 100MM devices generating 40k msgs / sec and be able to horizontally scale for future needs.

  • Leading Telco wanted to evaluate alternatives for a modern architecture and invited MongoDB to demonstrate fit.

  • Support multiple workload types including raw time series data and usage aggregates.

  • Support high write and read throughput.


Solution
  • Designed and developed an optimal data processingpipeline using Confluent Kafka, MongoDB, Kotlin and Spring Boot services deployed on OpenShift.

  • Demonstrated working solution at scale in 6 weeks timeframe.

  • Solution scalability and performance demonstrated not just Day 1 workload but also for Day 121.

  • Cost optimization and scaling strategies laid out.
“Client/PeerIslands Project Leader Testimonial Placeholder lorem ipsum dolor sit amet ri flor consetetur sadipscing elitr, sed diam nonumy eirmod temporil et in consetetur sadipscing elitr.”
Name
Company and Designation

Solution

  • Designed and developed an optimal data processingpipeline using Confluent Kafka, MongoDB, Kotlin and Spring Boot services deployed on OpenShift.
  • Demonstrated working solution at scale in 6 weeks timeframe.
  • Solution scalability and performance demonstrated not just Day 1 workload but also for Day 121.
  • Cost optimization and scaling strategies laid out.

Share this Article

Latest Case Studies