Cloud computing relies on cloud infrastructure keeping pace with its massive scale and unending increase in demand. However, the continued doubling of infrastructure performance generation after generation is becoming harder to sustain, particularly when coupled with the need for extremely high availability. Our mission is thus to invent infrastructure that can meet the scale, performance and availability goals of the future cloud. To this end, we leverage a key opportunity offered by the cloud – of stepping away from legacy concerns and developing new technologies for data centres.
Our work spans all layers of the cloud stack. Driven by concerns regarding continued CMOS scaling, our Optics for the Cloud program leverages and adapts bleeding-edge optical technologies for cloud resources, i.e., the compute, storage and network in the cloud. We are also developing Systems for the Cloud with a focus on hardware-software codesign to ensure that cloud applications will actually be able to the benefit from the improved performance and reliability that emerging hardware technologies have to offer. We believe that such vertical integration will be critical to drive innovation in the cloud space.
Our team is highly cross-disciplinary with computer scientists working hand-in-hand with physicists and partnering closely with product teams across Microsoft. Our researchers and engineers have diverse expertise across many areas of systems (distributed systems, storage, operating systems, networking, verification, hardware and electronics, etc.) and of optics (free-space, system-level and device-level). More broadly, we also collaborate with leading academic groups across the world to tackle big and open challenges facing next-generation cloud infrastructure.