IT@Intel Cloud Containers as a Service
Read this brief to see how Intel IT is implementing and using containers as a service.
Intel IT continually evaluates technology trends to ensure our infrastructure is efficiently meeting Intel’s business needs. Like most large enterprises, we transitioned from non-virtualized infrastructure to virtual machines (VMs) in the 2000s—improving resource utilization, reducing data center footprint and achieving better scalability. As we continue to refine our enterprise private cloud, VMs still provide benefit. However, the evolution of the cloud propelled the sheer pace of innovation everywhere. Constantly changing business needs resulted in the prominence of new design patterns (such as microservices) and the ability to scale solutions quickly anywhere becoming a necessity.
In 2013, Docker was released as an open-source engine that allows developers to package applications into containers and manage them with simple commands and automations. It was a major evolution in enterprise management. We were intrigued by the possibilities of using containers to enhance scalability, portability and efficiency.
Containers provide another level of architecture abstraction through operating system (OS) virtualization. This allows multiple applications to run their own separate processes while sharing a common OS, including resources like the CPU, memory and network I/O. Containers consist of an application and all its libraries and dependencies without needing a special hypervisor or guest OS. For this reason, containers offer greater density, faster boot times, and genuine portability, while consistently executing across multiple on-premises and cloud environments.
After evaluating Container as a Service (CaaS) trends, use cases, and solutions, we chose an open-source strategy aligned to the Cloud Native Computing Foundation (CNCF). This allows us to leverage the rapid innovation taking place among the more than 100,000 contributors to meet the constantly changing needs at Intel. Additional factors that we identified and analyzed across the technology providers included supplier maturity, market adoption, capability and innovation. We evaluated the suppliers’ approach to containers as either a “pure play” solution or a pivot from an existing Platform-as-a-Service (PaaS) architecture and determined the best fit for Intel’s enterprise needs. We also engaged peer IT organizations within the industry for peer-to-peer sharing and to learn about best practices and potential pitfalls.
To meet the varied needs of multiple business units, we wanted our CaaS platform to support two consumption models—a shared model and a dedicated model. Our CaaS platform’s agility supports both consumption models while maintaining the same roles and responsibilities for Intel IT. Application management belongs to the application/cluster owner in both models while the underlying infrastructure, software upgrades, monitoring, security, backups and training are managed by the platform team. This alleviates the infrastructure responsibilities and complexity for the application teams.
Multiple key platforms and capabilities at Intel are now hosted on CaaS, with project teams and users experiencing the benefits of increased agility, portability, efficiency, optimization and resiliency. Intel’s CaaS platform growth continues to accelerate. What began as a few dozen applications has expanded to cover a wide spectrum of containerized and deployed applications across Intel’s cloud enterprise.