The Nomad container orchestration alternative from Hashicorp and the Isovalent Cilium networking overlay for cloud-native apps are being combined by Cosmonic through the open source Netreap project.
According to Cosmonic’s head of infrastructure, Dan Norris, the company leverages both Nomad and Cilium in its platform-as-a-service (PaaS) environment for creating and deploying WebAssembly (Wasm)-based apps. Wasm is a portable binary instruction format for programming that executes in a memory-safe, sandboxed environment.
By integrating Nomad and Cilium, which is currently being developed alongside Kubernetes under the aegis of the Cloud Native Computing Foundation (CNCF), Netreap expands the capabilities of the Kubernetes container network interface (CNI). Only Kubernetes clusters were supported by Cilium in the past, however Cosmonic now supports Nomad as well. Kubernetes management may be a full-time job, which is a problem that many IT organisations are running into. Nomad offers a more manageable, lighter-weight option for orchestrating containers.
Although it’s unclear whether Nomad will catch on as a Kubernetes substitute, as more Nomad nodes are deployed in cloud-native IT environments, networking them becomes more important for businesses that have chosen to use Nomad. Some businesses are also using a combination of Kubernetes and Nomad nodes, which may require sharing network access.
The extended Berkley Packet Filter (eBPF) feature, which enables networking, storage, and security operations to be processed in a sandboxed environment that runs in the Linux kernel rather than in user space, is already utilised by Cilium. As a result, compared to traditional methods for building virtual networks, Cilium performs substantially better overall.
When integrating microservices operating on different pods across different clusters, cloud-native application environments need that degree of performance to drive networking. the development of application networking made possible by service meshes and proxy software operating over networks like Cilium.
Application networking has significant ramifications since it offers a chance to programmatically include network and security operations into DevOps procedures. Application networking offers a higher degree of abstraction for invoking those services without needing to rely on a network specialist, as opposed to requiring developers to know lower-level networking APIs.
IT organisations may need to review their organisational structure shortly. There may always be a need for dedicated networking specialists to manage the underlying physical network, but as network operations increasingly become an extension of a DevOps workflow, the networking services themselves will inevitably become more integrated with other methods of managing infrastructure-as-code (IaC).
No matter how application networking develops, it is evident that the rigidity that has long defined the provision of network services is vanishing as distributed application environments proliferate. Finding the delivery method that will cause the least amount of resistance for the network services that those latency-sensitive application services rely on presents a difficulty.