Federico Paolinelli

MetalLB is a load-balancer implementation for bare metal Kubernetes clusters, using standard routing protocols, including BGP. In the past year, the native go based BGP implementation has been replaced by an integrated FRRouting instance running in a container. In this talk I am going to briefly introduce how MetalLB leverages BGP to implement Kubernetes services, describe the integration mechanism with FRR and how it made the MetalLB evolution faster and easier. I'll finally explain how the MetalLB CI works, and how we leveraged FRR to mimic various network topologies in order to extend number of scenarios we want to cover.

Being the maintainer of a fairly active oss project (MetalLB) over the past year, I reviewed a substantial amount of contributions. During this process, I identified a set of recurring idioms and patterns that less experienced contributors keep missing, making the codebase harder to read and to maintain.

In this talk I will describe what cognitive load is and why it matters, and provide a way to reduce it via a set of quick and easy recipes. Using this set of actionable recipes the audience will be able to drastically improve the quality of their Go code with relatively low effort.