Website: Microsoft Research Summit 2021 (opens in new tab)
Future of Cloud Networking at Microsoft Research Summit
For 30 years, our research community at Microsoft has worked across disciplines, institutions, and geographies to envision and realize the promise of new technologies for Microsoft and for society. Today, we’re inviting the global science and technology community to continue this exploration—because ensuring that future advancements benefit everyone is up to all of us.
Microsoft Research has been at the forefront of network innovation for the cloud for over a decade. Our work, in partnership with Azure, spans all aspects of the network stack, along with network management and verification. Our research has shipped in various parts of the Azure stack and has won several awards. In this session, you will learn about some of these innovations and current problems we’re working to solve, from researchers in Microsoft Research and from leaders in Azure.
| Introduction & Keynote | |
|---|---|
| Ranveer Chandra (Head, Networking Research, MSR Redmond) Dave Maltz (CVP, Azure Networking) |
|
| Topics | Speakers |
| PHY | Hitesh Ballani (Sr. Principal Researcher, Microsoft Research Cambridge) |
| Switches/NICs | Alec Wolman (Sr. Principal Researcher, Microsoft Research) Lihua Yuan (Partner Engineer Manager, Microsoft) Daniel Firestone (Partner Engineer Manager, Microsoft) |
| Net Management | Behnaz Arzani (Senior Researcher, Microsoft Research) Ranjita Bhagwan (Sr. Principal Researcher, Microsoft Research) |
| Network Verification | Andrey Rybalchenko (Sr. Principal Researcher, Microsoft Research Cambridge) George Varghese (Professor, UCLA) |
| WAN | Jitu Padhye (moderator, Partner Engineer Manager, Microsoft) Sylvia Ratnasami (Professor, UC Berkeley) Erica Lan (CVP, Azure Networking) |
| Apps | Venkat Padmanabhan (Deputy Managing Director, Microsoft Research India) Zhixiong Niu (Senior Researcher, Microsoft Research) Francis Yan (Senior Researcher, Microsoft Research) |
| Software Defined Networks | Deepak Bansal (CVP, Azure Networking) Srikanth Kandula (Sr. Principal Researcher, Microsoft Research Redmond) |
| Closing Keynote and remarks | |
| Victor Bahl (Technical Fellow, CTO, Azure for Operators) Lidong Zhou (Distinguished Scientist, Microsoft Research) |
|
This track will take place in three broadcast regions:
- October 18, 8:00 PM–11:00 PM China Standard Time (UTC+8),
- October 19, 2:00 AM–5:00 AM British Summer Time (UTC+1), and
- October 19, 9:00 AM–12:00 PM Pacific Time (UTC-7)
Once registered, you can view the session scheduler, select the broadcast region nearest you, and add the Research for Industry track to your event schedule. Learn more about Microsoft Research Summit (opens in new tab).
Track sessions
| Session | Description |
| Opening remarks |
Speaker: |
| (Keynote) Scaling Challenges in Cloud Networking | The promise of cloud computing is unlimited resources that customers can put towards solving the largest problems they face. Yet realizing that vision requires solving myriad problems in physical systems, distributed systems, and data analytics. This talk will examine some of those challenges and approaches toward solutions.
Speaker: |
| Lightning Talk: Optical Networking | Emerging cloud workloads like large-scale AI and new system drivers like hardware accelerators and resource disaggregation will require significantly higher bandwidth, lower latency and higher reliability than today’s applications. It will be very hard to meet these requirements with mainstream network technologies that are starting to be impacted by the slowdown of Moore’s Law. This perfect storm rules out incremental improvements as the network infrastructure needs to be at least an order of magnitude more efficient to cater to future cloud requirements. In this lightening talk, Hitesh Ballani, Researcher at Microsoft, will outline this need for disruptive change and will also explain how Microsoft Research is betting on optical innovation to create, from the ground up, new network technologies for both switching and transmission that could potentially offer the required step-change in cloud network performance.
Speaker: |
| Panel: Blurring the gap between software and hardware in networks | Over the years we have made and continue to make investments on exerting fuller software control on traditionally opaque hardware such as switches and routers as well as on moving traditionally software functionality into hardware accelerators. In this panel, we bring together product and research perspectives on the case for these efforts, where we are and what to expect in the near and far future.
Speakers: |
| Lightning Talk: Revisiting data center management: moving towards self-managed data center networks | Meeting service level agreements (SLAs) is critical for data center operators, and the health of their networks will determine whether an operator succeeds or fails at this task. Despite the many advances in verification, testing, and network diagnosis and monitoring, many engineering hours are still spent on managing data center networks, diagnosing problems when they happen, and mitigating impact on customers when the network is unhealthy. Automation can help reduce errors and increase efficiency. In this session, you can learn about the various Microsoft Research workstreams that focus on automating network management and providing operators with more sophisticated risk assessment, monitoring, diagnosis, and mitigation tools, all of which move us closer to self-managed data center networks.
Speaker: |
| Lightning Talk: Data-driven configuration management for cloud services |
Speaker: |
| Panel: Network verification: Cloud first, what’s next? | Delivering a reliable network to customers is top business priority. Providers can prevent network misconfigurations from degrading reliability by verifying every network configuration change before its deployment. This is a hard problem to solve, given the scale and complexity of today’s networks. Join Microsoft researcher Andrey Rybalchenko, from the Cloud Infrastructure group at Microsoft Research Cambridge, UK, together with UCLA Professor George Varghese, to discuss the progress in network verification research and its applications in industry. They will review some recent advances and unsolved problems and offer a preview of where the research is heading next.
Speakers: |
| Panel: Software Defined WAN’s | The past decade has witnessed exponential growth in cloud computing market, and the size and complexity of cloud service provider’s backbones has increased accordingly. In this fireside chat, Erica Lan, leader of Microsoft Azure networking team, and Professor Sylvia Ratnasamy of UC Berkeley discuss the future of backbone networks for large cloud service providers. Both Erica and Sylvia were key players in this paradigm shift; with Erica helming massive Azure WAN and Sylvia contributing many research ideas, and co-founding Nefeli Networks. During this discussion, we will ask both leaders to peer into the crystal ball and tell us what they believe the next ten years will bring.
Speakers: |
| Fireside Chat: Networking for Cloud Apps | Today, applications deployed in the cloud or at the edge provide services to end users over a heterogeneous array of networks, from high-speed wired links to fragile wireless communications. Data-driven techniques, including machine learning (ML), can enable the realistic evaluation of networked applications using simulators and emulators, and may equip such applications with better adaptability and versatility in these diverse scenarios. However, achieving good performance in practice demands thoughtfully designed ML algorithms as well as diverse learning environments. Join us for a discussion on this timely topic with Microsoft researchers Francis Y. Yan and Zhixiong Niu, moderated by Venkat Padmanabhan. During this discussion, they will introduce: OpenNetLab, an open distributed platform of heterogeneous nodes, established to promote data-driven networking research, and the iBox (Internet on a Box) data-driven simulator; MMSys ’21 grand challenge on bandwidth estimation for real-time video, showcasing OpenNetLab’s capability to foster ML-based network algorithms; and a related cross-lab research project which aims to leverage reinforcement learning to optimize quality of experience for videoconferencing users.
Speakers: |
| Fireside Chat: Smart network pipes unleashing new opportunities | Network device programmability and capability are growing rapidly, driven by innovations like programmable data planes and powerful control planes with cost effective ARM cores. This has enabled more extensive packet processing in the networks. Such smart networks are unleashing new opportunities, starting with cloud data center and edge networks. In cloud data center, network processing can offload complex functions like access control, load balancing, encryption, and policy application from the servers to the network, freeing up server capacity to run application code/business logic. On edge networks, particularly 5G, smart networks can provide video compression and detection, alerting, data aggregation and other functions close to data source generation. This eliminates the need to haul the traffic back to the data center and reduces the need to deploy lots of compute capacity at the edge. In this fireside chat, Deepak Bansal will discuss the trend towards smart pipes and the new scenarios it is unleashing.
Speakers: |
| (Keynote) The Future: Converging the Cloud & Telecommunications Infrastructures | The emergence of 5G has created an unexpected, unprecedented shift for telecommunications and the link between telecommunications and computing. In this talk, Victor Bahl will describe the scientific advances bringing us to where we are today with 5G, and then cast an eye to the future in sharing with the audience a vision for where things are going with telecommunications, including key enablers and potential surprises on the horizon. This discussion will set the context for describing the opportunity ahead for the academic community the next several years, and beyond, as we stay at the forefront of innovation in telecommunications, with innovations with the cloud and edge. We will next move into near-term strategy with our standing up of Azure of Operators (AFO), which is aimed at bringing telecommunications industry to Azure as an integral component of its global infrastructure. This talk will describe the motivation and vision that led to the creation of AFO, its mission, and the significant technical and scientific challenges, which when overcome will lead to the inevitable convergence of two massive industries and new opportunities for Universities and research institutes.
Speaker: |
| Closing Remarks |
Speaker: |