SkyAlert Helps Save Lives from Natural Disasters Using AWS Solutions

SkyAlert is an innovative technology company that quickly alerts millions of people living in earthquake-prone areas, promoting a culture of prevention against natural disasters. In order to provide customers—both businesses and individuals—with the right tools to protect them during earthquakes, SkyAlert migrated its infrastructure from Microsoft Azure to Amazon Web Services (AWS). After implementing its Internet of Things (IoT) solution to run on AWS and its efficient alert service, SkyAlert scales quickly and can send millions of messages in a few seconds, helping to save lives in the event of earthquakes.

SkyAlert offers an early warning system for earthquakes in Mexico and some western regions of the United States. Serving businesses and individuals, SkyAlert currently has more than three million registered users who rely on its alerts in the event of an earthquake. The emergency nature of SkyAlert’s service requires infrastructure that can dynamically scale so that messages are sent and arrive on time so that users can take appropriate safety actions. In the case of IoT, smart devices must be activated well in advance to help avoid catastrophes.

Francisco Catalá, chief technology officer (CTO) of SkyAlert, explains the application’s particular needs: “On a day-to-day basis, the application has low traffic. However, when there is an earthquake warning, traffic grows exponentially and very quickly, which requires the application to scale quickly to meet demand.”

Before migrating to AWS, SkyAlert’s architecture required 20 virtual machines to work simultaneously and took hours to scale—millions of message alerts generated by an earthquake warning resulted in bottlenecks and alert delays. “Even a minute delay in delivering an earthquake alert to a user is crucial in the case of disaster prevention,” says Mario Ferreira, principal architect of SkyAlert.

SkyAlert needed to invest in modernizing the infrastructure behind its services. “Seismic activity in Mexico never decreases and will always be a concern, so it is very important to have efficient prevention technology,” Catalá explains.
 

In 2018, SkyAlert decided to migrate and used AWS Partner Cloudup to make a proof of concept to migrate machines to containers. Using AWS Lambda, AWS Fargate, and Amazon Pinpoint to scale automatically, SkyAlert could deliver messages to users quickly.

Using Amazon Elastic Container Service (Amazon ECS) and Amazon Elastic Kubernetes Service (Amazon EKS) to build the framework, which both help manage and scale applications in the cloud, SkyAlert was able to segment the delivery of messages as needed. For example, the company configures messages to users based on geolocation and can interrupt or modify messages during the alert process, something previously impossible with Azure. SkyAlert also uses Amazon CloudWatch (CloudWatch), an observability service that provides data and actionable insights, to monitor devices on the IoT network.

Today, SkyAlert is on version four of its application and has an alert system composed of more than 120 sensors installed in strategic geographical locations in Mexico and North America, all of which are integrated with AWS for analytics, metrics, and push notification functions. In addition, SkyAlert uses AWS for its IoT solution to activate equipment that helps keep people and facilities safe. For example, SkyAlert can control the automatic opening of gates for firefighter access or close gas and electrical systems to prevent fires. SkyAlert completely redesigned its product to work more seamlessly with cloud services to provide the same solutions for businesses and individuals alike.
 

The June 2020 earthquake in Mexico provided proof that the new cloud-based structure best supports SkyAlert by delivering nearly 2 million notifications. “The whole system worked perfectly. Lambda functions scaled from 1 to 41 in less than 10 seconds,” explains Ferreira. He points out that downtime for scaling, which previously took 40–50 minutes, does not exist today with Fargate, a serverless pay-as-you-go compute engine. “The system scales by itself in a matter of seconds,” Ferreira explains.

In the previous SkyAlert warning structure, alert messages often took up to 20 minutes to arrive, but the current cloud-based structure alerts millions of users within 8 seconds. “Using AWS services allows us to segment alerts and quickly scale to help save lives,” says Catalá, adding that users can know where and for how long they will feel the earthquake.

During the June 2020 earthquake, SkyAlert gave users more than 120 seconds to take safety measures. “We were able to communicate to our users up to 20 seconds before other alerts fired their horns on the streets,” Ferreira celebrates. “We also received a lot of encouraging comments from people saying that the app saved lives.”

The benefits of migrating to AWS goes beyond technical efficiency. Previously, SkyAlert had 20 machines running around-the-clock. With AWS, containers allow the company to scale dynamically when needed, reducing the number of necessary machines to as few as seven. According to Ferreira, the previous monthly costs amounted to $7,000 USD; using AWS, costs went down to $4,000 USD, with better performance, larger and more modern infrastructure, and tools for data analysis and tracking the delivery and opening of messages.

In short, migrating to AWS allowed SkyAlert to become a leader in earthquake warning solutions in Mexico. “No one else can deliver the information with the intensity and efficiency that the user needs,” says Catalá.
 

Despite already achieving remarkable results and recognition from users, SkyAlert believes there are still many opportunities for improvement. It has a goal to send alerts in less than 4 seconds. “We can still promote a lot of improvements. Beyond improving what we already have, we want to implement the entire IoT suite of AWS products to not only improve the line of business, but also completely revitalize it,” Ferreira says.

The company must also invest in the backend and in analytics. “We still have a lot to do to optimize the monitoring issue within our cloud environment. We want to implement more CloudWatch services and be much more precise. SkyAlert is growing every day, and we need to be in control. Another forthcoming improvement is to implement analytics with Amazon Kinesis to deliver exactly what the user wants, and measure the performance and usage of the solutions closely to improve products even further,” explains Catalá.

For Ferreira, it comes down to support team quality and innovation: “The quality of the AWS team differentiates them from others. The assistance from AWS architects and technical support—electronic or face-to-face—is better, more efficient, and more agile.”