Globe Group is one of the largest telecommunications companies in the Philippines, with almost 87 million customers. A cloud-native since its 2014 launch, 70-80% of its workload is on AWS. Globe’s focus has always been rapid digital delivery.
With 800 applications across 7,000 servers and 5.8 petabytes of storage on public and private clouds, a recent incident galvanized Globe to re-energize its cloud usage and innovate faster.
Presenting at MongoDB.local Singapore, Ricky Mohan, Globe Telecom’s Head of Cloud Enablement and Service Operations, told the story.
On December 25th, 2022, Ricky got notification of what appeared to be a distributed denial-of-service (DDoS) attack. Customers in huge numbers were attempting to visit Globe’s website and apply to register their SIM cards, as they were required by a new government mandate to do. If they failed within a given time, they’d lose their phone number.
But Globe hadn’t designed the system to have millions of customers accessing its call systems downstream. Each time they solved one point of failure, another would fail behind it. The incident lasted over two days.
Emerging from the crisis, Ricky and the team realized that most of Globe’s legacy systems had backend synchronous dependency, so that only once one operational task was completed could the next be tackled. They began to rethink how to ‘do digital at Globe.'
The team started small, with an auto-capturing site for SIM-card-only registration. They wanted the customer journey to take a zero-negative path, meaning that if a particular system failed, instead of customers receiving a ‘try again later’ message, Globe could capture the order and process it themselves. The aims were to:
- never go offline
- Zero-touch provisioning
- reduce human involvement
To meet the challenge, Globe formed a cross-functional engineering group with MongoDB –MongoDB Atlas was already supporting several of Globes products and services – as a key partner.
Already running on AWS, the team chose Google Cloud Platform (GCP) for the project from a multi-cloud perspective. “It benefits us to have healthy competition between multiple partners,” explains Ricky, “and MongoDB Atlas being multi-cloud by default allows us to use data across multiple cloud vendors.” The infrastructure was configured with a multi-cloud platform engineering mindset, creating patterns that could be re-used by developer teams so they could be quicker to realize business value.
Typically, at a large-scale enterprise, a project of such complexity, involving SAP, provisioning systems, and Operational Support Systems (OSS) and Business Support Systems (BSS) layers, would take at least nine months to deliver. The Globe team nicknamed their effort, ‘Project Impossible.’ Here’s why:
- 4 months to delivery
- Nearly 90% reduction in opertational running costs
- 70% reduction in human involvement to process orders
“We were able to deliver so quickly because of the technology choices we made to reduce the work of developers, and MongoDB played a major part in that,” says Ricky.
