Computer Science Special Lecture November 2, 2018 11:00am — 12:00pm Location: McWilliams Classroom 4303 - Gates Hillman Centers Speaker: DANIEL FIRESTONE, Principal Tech Lead and Group Manager https://www.microsoft.com/en-us/research/people/fstone/ Accelerating Software Defined Networking in Microsoft Azure Speaker: Daniel Firestone, Principal Tech Lead and Group Manager Location: GHC 4303 Accelerating Software Defined Networking in Microsoft Azure Modern cloud architectures rely on each server running a software-defined networking stack to implement policy and virtualization. However, these networking stacks are becoming increasingly complex as features are added and as network speeds increase. Running these stacks on CPU cores takes away processing power from VMs, increasing the cost of running cloud services, and adding latency and variability to network performance, which has led us to invest in technologies to accelerate SDN in NIC hardware without sacrificing software programmability. We’ll present Azure Accelerated Networking, which uses FPGA-based Azure SmartNICs to provide direct host-bypass networking to VMs while implementing SDN policy, and describe our evolution from soft switching to programmable SmartNICs, comparing what we can do in switches vs hosts, and different SmartNIC approaches. We’ll also talk about our related work on Cloud DPDK, with which we’ve enabled accelerated NFV scenarios inside Azure VMs on top of SmartNICs.—Daniel Firestone is the principal tech lead and group manager for the Azure Host Networking team. His team builds the Azure virtual switch, which serves as the datapath for Azure virtual networks, load balancers, security groups, QoS, and more, as well as SmartNIC, the Azure platform for offloading host network functions to reconfigurable FPGA hardware, and the RDMA accelerated transport layer behind Azure Storage. Before Azure, Daniel did his undergraduate studies at MIT.Faculty Host: Justine SherryGuest lecture in 15-829 (Programmable Networks) For More Information: Appointments