Large-scale layer-2 Ethernet networks are needed for important future and current applications and services including: metro Ethernet, wide area Ethernet, data center networks, cyber-physical systems, and large data processing. However Ethernet bridging was designed for small local area networks and suffers scalability and resiliency problems for large networks. I will present the architecture and protocols of ROME, a layer-2 network designed to be backwards compatible with Ethernet and scalable to tens of thousands of switches and millions of end hosts. We first design a scalable greedy routing protocol, Multi-hop Delaunay Triangulation (MDT) routing, for delivery guarantee on any connectivity graph with arbitrary node coordinates. To achieve near-optimal routing path for greedy routing, we then present the first layer-2 virtual positioning protocol, Virtual Position on Delaunay (VPoD). We then design a stateless multicast protocol, to support group communication such as VLAN while improving switch memory scalability. To achieve efficient host discovery, we present a novel distributed hash table, Delaunay DHT (D²HT). ROME also includes routing and host discovery protocols for a hierarchical network. ROME protocols completely eliminate broadcast. Extensive experimental results show that ROME protocols are efficient and scalable to metropolitan size. Furthermore, ROME protocols are highly resilient to network dynamics. The routing latency of ROME is only slightly higher than shortest-path latency. / text