decorative pattern
decorative pattern

Graviton Migration Case Study

Bria.ai leveraged Opsfleet expertise to make the shift a seamless process.

Let's Talk

The Client

Bria.ai is a fast-growing startup specializing in AI-powered image processing and optimization solutions. The company leverages advanced computer vision and machine learning algorithms to develop solutions to automate various image processing tasks, such as object recognition, image classification, and content moderation.

Region
Israel
Industry
AI Image Processing
Main Technologies
AWS, Kubernetes, EKS, EC2, Graviton
Services
DevOps-As-A-Service
Date of Project
Problem

The Challenge

As a company that specializes in AI-powered Image processing, Bria.ai's business has rapidly grown over the past couple of years, reaching a massive scale as they onboard more and more customers.

To facilitate their scale and to make sure they are doing so efficiently as possible, Bria's engineering team realized that their existing EKS Node Groups that are using intel CPUs were either too expensive or, in some cases, not delivering the required performance to meet their SLA.

The ideal solution for Bria was to migrate their EKS workloads to EC2 Graviton Instances that are ARM-based and provide better performance at a better price than traditional EC2 Instances.
However, successfully migrating EKS applications that were built for traditional Intel-based EC2 to ARM-based Graviton requires some level of expertise and DevOps resources.

Since Opsfleet was the partner that helped Bria set up their initial infrastructure, it was only natural that they would select Opsfleet once again as they were very well acquainted with their environment and that would make the migration a seamless process.

What we’ve done

The Solution

Since OpsFleet is an advanced AWS Partner, we had access to AWS resources and funding to kick off the migration project.

The implementation phase consisted of 3 main milestones:

  • Identify the dependencies that needed to be modified and perform those changes, including tests to validate functionality
  • Update the CI build process to build images that support two architectures
  • Add Graviton node groups to the infrastructure using IaC and update CD pipelines

The biggest time-wasting effort for Bria's developers was tinkering with the code libraries, and that's where Opsfleet came in to manage and streamline the entire process.

After successfully executing sanity and regression tests on Bria's Development environment, the updates were applied to their production environment as well.

Due to the seamlessness nature of the migration process and the high level of trust Bria had in the team, Opsfleet expanded Bria's Graviton deployment to RDS Instances as well, which were originally out of the scope of this project.

Results

The Outcome

After collaborating with Opsfleet, Bria's engineering team successfully shifted their EKS applications and RDS instances to Graviton-based instances and now they benefit from better DB and application performance, and a lower EC2 bill.

Opsfleet not only completed the migration at the highest standard, but saved Bria's engineers plenty of time and effort in a long and daunting migration process.

The experience we got from Opsfleet is that our team just pushes the button and it works on Graviton.

Bar Fingerman
Head of Engineering
quote decorative illustration

Next Projects

Imagen Standardized their AI Cloud Infrastructure and scaled from dozens to 10K+ users

Learn how Opsfleet helped Imagen AI build and scale their application on AWS with Fractional DevOps

Migrating to Kubernetes for Scalability and Flexibility

Optibus, a SaaS platform that plans and schedules mass-transportation through AI and optimization algorithms, faced challenges in scaling their CPU-intensive system due to cumbersome scaling cycles and custom Python code.

Opsfleet Helps Bria.ai Migrate to Kubernetes and Simplify Their Development Process

As the company grew and started onboarding its first enterprise clients, the team realized that the existing EC2-based infrastructure didn’t meet the demands of large enterprise clients who expect high availability and production-grade reliability from their 3rd party software providers.