IT Outposts' Role in DreamFlare's AI Streaming Revolution

DreamFlare is a groundbreaking streaming service that pushes the boundaries of digital entertainment. It presents AI-generated animated videos and offers filmmakers a space to share and monetize their content. The platform features various video formats, including flips, which bring comics to life, and spins, which allow viewers to interact with the story and choose different paths while watching it (much like the interactive Black Mirror episode).

DreamFlare made a big splash when it launched, with articles in major media outlets like Forbes, TechCrunch, and BusinessWire. Furthermore, the project is guided by a former Google employee and a documentary filmmaker. So, from the start, DreamFlare was designed to bring together high-tech ideas and the art of storytelling.

Jump straight to a key chapter

Project Description
Provided Services
Work Agenda
Tasks and Challenges

Solutions
Results
DevOps Tech Stack

Project Description

When we first met with the DreamFlare team, their project was already in development. They needed DevOps expertise for a successful launch, and the main requirement was to build a stable infrastructure.

As an innovative company, DreamFlare has ambitious goals. However, at that point, the platform wasn’t ready for all the innovations they had in mind. There was still a lot of work to be done before our team could implement everything our clients envisioned.

Our initial task was to bridge that gap — to create the infrastructure that would meet current needs while laying the foundation for the future.

Provided Services

DevOps Services

SRE services

Log management and monitoring
Incident management
Release management
Performance management

Operations managed services

Capacity planning
Disaster recovery as a service

DevSecOps services

Security assessment
Access control and identity management

Work Agenda

Client

DreamFlare, a streaming service for interactive AI-generated content

Location

USA

Technical team

2 DevOps engineers
Project manager

Project timeframe

March 2024 - ongoing

Project goals

Implement a stable version control system

Protect the brand identity and manage multiple domain variations

Establish a robust CI/CD pipeline tailored to the client's Amazon-centric environment

Secure the database while maintaining necessary access for the development team

Set up comprehensive monitoring and alerting for the platform

Ensure the infrastructure can handle the traffic

Create a secure pre-release testing environment with controlled access

Implement IaC practices for better management and scalability

Tasks and Challenges

Chaotic version control

One of the biggest issues was that our clients relied too much on the main branch. Team members created separate branches for their work, but everything was pushed straight into the main branch. This became increasingly problematic as the project grew in complexity.

No CI/CD

The clients’ existing deployment workflow significantly impaired the team's productivity and jeopardized the entire project. The more complex the platform became, the more urgent the need for a more structured and reliable deployment process became. The DreamFlare team needed a new deployment strategy without disrupting ongoing development work.

No comprehensive monitoring and alerting

Without proper monitoring, the team was flying blind. They had no real-time insight into the health of their systems, so it was impossible to proactively identify and fix issues before they grew into major problems. The lack of insight meant that performance bottlenecks, resource constraints, and potential security threats could go unnoticed until they impacted users.

Provide selective access to the website before the official release

The DreamFlare team also requested to keep their website accessible from the internet while limiting access to authorized personnel only. Sure, we could have implemented a VPN-based solution to restrict IP access, but it wasn't sufficient for thorough testing. The clients needed to simulate real-world conditions as closely as possible, which meant testing every aspect of the website, including all potential user scenarios and edge cases.

For this, we designed a basic authentication system that would lock down the entire website, including the home page, to the public, so only team members with the correct credentials could access and test the platform.

Protect brand identity and online presence

Our clients also needed to prevent the unauthorized use of their brand name across various domain extensions and similar-sounding domains. This included protection against cybersquatting, where others can register domains similar to DreamFlare for malicious intent. They also wanted to protect against phishing attempts, where similar domains can be used to trick users into revealing sensitive information and damage DreamFlare's reputation.

Fortify database access in a private network against potential breaches

The database is the heart of the streaming platform, where all kinds of sensitive information is stored — user data, content details, and more. Not only did it need to be protected from unauthorized access, but it also had to be accessible to engineers so they could work efficiently with the data.

This balance was critical — a data breach could have serious consequences, but overly restrictive access could hinder the team's ability to respond quickly to issues.

Ensure infrastructure readiness for launch day traffic

As DreamFlare approached its highly anticipated launch, they wanted to make sure the infrastructure could handle the launch day traffic. The clients had set a target for the expected number of users and needed to verify their systems were up to the task.

Keep infrastructure in check

As the DreamFlare platform expanded, the clients required a way to track all the changes to their systems. It wasn't just about knowing what had changed but also about being able to reliably restore these changes.

So, without infrastructure as code (IaC), they would have to remember every small change to a complex system for months or even years. The clients would also have to manually rebuild everything from scratch instead of recreating their entire infrastructure with just a few clicks.

Contacts

Contact IT Outposts to discover how we can build a robust and scalable infrastructure tailored to the unique needs of AI-driven innovations, just like we did for DreamFlare. Whether you’re looking to streamline your AI development pipeline, secure your platform, or ensure seamless scalability, our expertise in DevOps and cloud solutions is the perfect fit for your AI startup.

Discuss your project

Solutions

1. Add a staging environment

First of all, we’ve introduced a staging environment. Instead of pushing updates directly to production, the DreamFlare team can now push all changes to the staging environment first. This is an intermediate step that allows them to thoroughly review and test the updates before they go live.

2. Build a robust CI/CD pipeline

Our team typically configures CI/CD pipelines with GitHub Actions — a versatile tool that we've relied on for many projects. However, this situation required a different approach. With about 90% of the clients' services coming from Amazon, it made sense to adapt the CI/CD process to the existing ecosystem.

So, we created a CI/CD solution with AWS services. Our engineers configured AWS CodeBuild to manage the build process, while AWS CodePipeline was set up to orchestrate deployment.

3. Implement monitoring and alerting using Amazon CloudWatch and EventBridge

The IT Outposts team built a robust monitoring system to keep an eye on the DreamFlare platform. We used Amazon's native tool, CloudWatch, to handle all their monitoring needs.

For prompt alerting, our engineers also integrated the CI/CD pipeline with EventBridge, another Amazon tool. When a deployment finishes, it automatically sends out an email — a "success" message if everything went smoothly or a "failure" alert if something went wrong.

To make things even more convenient, we then configured AWS Chatbot, another Amazon service. It allows for sending notifications directly to the clients’ Slack channel.

4. Establish basic authentication using AWS CloudFront and a Lambda function

To implement the basic authentication requirement, our team found a clever solution using, again, an AWS service. Particularly, we set up CloudFront, a content delivery network service, and coupled it with a Lambda function.

When a user tries to access the website for the first time, CloudFront triggers the Lambda function that handles the authentication process. As soon as the user successfully logs in, the system generates a token. So, instead of prompting the user to log in each time, it adds this token to the user's cache. This means that after the first login, the user can move freely around the website without having to constantly re-enter their login details.

5. Set up redirects

To protect the DreamFlare brand identity, our clients purchased multiple domain variations, such as DreamFlare.com, DreamFlare.co, and even DreamFlair.

To handle these extra domains, we created a redirect system using Amazon S3 buckets. Our engineers essentially developed an empty, static website whose sole purpose was to redirect traffic to the main domain. Next, we moved all their name servers from GoDaddy to Amazon's Route 53 service. This migration brought all their domain management under one roof.

Of course, security was a top priority, too. Our team set up SSL certificates to ensure encrypted connections. We also implemented a web application firewall that acts as a security guard for web applications, protecting against common web exploits.

6. Configure the database in a private network within the Amazon VPC

In addition, our engineers configured the database in a private network, adding an important layer of security. To provide the necessary access to the database, we set up bastion hosts — if someone needs to access the database, they must first connect to these bastion hosts, which then provide a secure path to the database.

Knowing that this setup might be new to some DreamFlare team members, the IT Outposts team held training sessions to make sure everyone understood how to use this new system. Our engineers also showed initiative and looked closely at the tools our customers were already using. They wanted to find ways to integrate the new system with the existing tools and make the transition as smooth as possible.

7. Conduct load testing, analyze results, and adjust the infrastructure accordingly

To ensure the clients’ infrastructure could handle the expected load at launch, we started by conducting load testing to simulate heavy user traffic and see how the system performs under pressure.

The results of these tests were revealing. They showed that to handle the expected traffic, some additional work was needed. Specifically, we had to allocate more resources to the infrastructure and adjust the autoscaling settings to provide more capacity.

Recognizing that things don't always go as planned, our engineers also developed a contingency plan, outlining steps to take if the traffic exceeded their capacity and the autoscaling couldn't keep up fast enough. This helped us ensure a smooth launch for DreamFlare's platform.

8. Adopt IaC practices

A crucial aspect of our approach was the use of Terraform for infrastructure management — everything we do is always described as code.

And even when we needed to test something quickly, like a Lambda function, we didn't leave it as a manual configuration. Our engineers might set it up manually for initial testing, but once they confirmed it worked, they went back and described it in Terraform code. This approach extended to all aspects of the DreamFlare’s infrastructure. Every detail, each zone, and almost every record was defined in Terraform.

The beauty of this method is that it made changes consistent and trackable. Now, when our clients need to make updates, they can just modify the Terraform code and apply these changes.

Results

Launch readiness. Our engagement helped DreamFlare successfully go live on and welcome its first users on time.

Enhanced brand protection. The domain management strategy secured DreamFlare's online presence, preventing traffic loss and maintaining customer trust.

Accelerated time-to-market. The improved CI/CD pipeline and staging environment shortened the time it took to test and deploy new features, allowing DreamFlare to innovate faster than the competition.

Stable infrastructure built to grow. Most importantly, we provided DreamFlare with a strong foundation to build on. They can now start bringing more of their creative ideas to life.