Labelbox•April 1, 2024
The dev corner: streamlining deployments at Labelbox with ArgoCD and GitOps
![](https://labelbox.ghost.io/blog/content/images/2024/06/YT-Thumbnails--29--1.jpg)
At Labelbox, we're constantly innovating to improve our development and deployment processes. Recently, we transitioned from a traditional continuous delivery (CD) pipeline to a GitOps approach using ArgoCD. This shift has brought significant advantages in terms of automation, collaboration, and overall deployment reliability.
We outgrew traditional CD (CIOps)
Our previous CD pipeline involved manual configuration management and scripting, which became cumbersome as our codebase grew. This approach also posed challenges in:
- Version control: Tracking changes and rollbacks became complex.
- Team & environments: As the number of team members, environments, and services we operate grew, it was difficult to know what was running at any given time.
- Push approach: Our traditional CI/CD system required credentials for each target environment, and every application needed its own pipeline.
The power of GitOps with ArgoCD
A GitOps approach, using ArgoCD, offered a compelling solution. Here's how they've transformed our deployments:
- Declarative configuration: We define the desired state of our infrastructure and applications in Git. ArgoCD responds to changes in Git and works to realize that state in our Kubernetes clusters.
- Automated deployments: ArgoCD watches for changes in our Git repository and triggers deployments automatically, streamlining the process and minimizing human error. Leveraging release branches, we can update several Development, Staging, and QA clusters in tandem.
Key benefits of our move to ArgoCD and GitOps:
- Reduced deployment Time: Automated deployments have significantly reduced the time it takes to implement new features and bug fixes in production.
- Improved reliability: Declarative configuration and GitOps workflows minimize errors and ensure consistent deployments. In addition, it surfaces issues more quickly.
- Stronger collaboration: The Git-based approach fosters better communication and collaboration between development and operations teams.
- Scalability: As we continue to grow, ArgoCD's architecture allows us to manage deployments across multiple Kubernetes clusters and environments easily.
- Security and reliability: Declarative configuration minimizes manual errors and inconsistencies, leading to more reliable deployments. Additionally, Infrastructure as Code practices promote consistent and secure deployments.
- Auditability and accountability: The Git commit history provides a natural audit log of what changes were made to application configuration, when they were made, and by whom. Combined with ArgoCD-emitted Kubernetes Events, we have greatly improved our traceability.
Conclusion
By embracing ArgoCD and GitOps, Labelbox has unlocked a new era of streamlined deployments. From automated rollouts and improved collaboration to enhanced security and scalability, this approach empowers our teams to focus on innovation and deliver exceptional value to our users.
If you’re looking to modernize your deployment process and leverage the power of GitOps, ArgoCD is definitely worth exploring.
About Labebox
Labelbox is the leading data-centric AI platform for building intelligent applications. We enable teams to capitalize on the latest advances in generative AI and LLMs by injecting these systems with the right degree of human supervision and automation. Whether you're building AI products using LLMs that require human fine-tuning or applying AI to reduce time in manually-intensive tasks, Labelbox empowers you to do so effectively and quickly.
Our platform is trusted by Fortune 500 enterprises including Walmart, Procter & Gamble, Genentech, and Adobe, as well as hundreds of leading AI teams. We're transforming industries from insurance and retail to manufacturing, robotics, and healthcare.
Interested in opportunities to build human-aligned AI systems?
We're always looking for talented individuals to help us push the boundaries of AI development.