VMware was proud to host the first ever future:net conference attended by approximately 160 senior network architects, principle engineers, CTO and master practitioners of cloud technology. This invitation-only event included speakers from eBay, Microsoft, IBM, Columbia Sportswear, Thomson Reuters among others presented on the transformation of networking driven by new technologies and initiatives.
We hope you found these two days to be valuable and informative and you had the opportunity to meet and interact with your peers. The session replays are now available, just click on each Track Title in the agenda below to view.
future:net is a complimentary event by invitation only, and invitations are non- transferrable. Airfare and lodging are not included.
Networking has changed a lot in the last ten years. One big change is the creation of whole new abstractions for controlling and programming networks, allowing us to borrow decades of software engineering from our colleagues. Nowadays, people who own and operate the largest networks take it for granted that they can specify, write, buy, commission or download software to control their network. Are we done? In the talk, I'll point out trends happening now that I think lay the foundations for the next leap forwarding in networking.
The awesome ability of cloud-scale networks to deliver tremendous bandwidth between over 100,000 servers at high availability and very low cost stems from their use of broad layers of cheap switches in highly redundant topologies like a Clos or Fat-Tree. However, the sheer scale of such networks means that “gray” failures, in which switches stop correctly forwarding some packets while still forwarding others, become common. These gray failures must be found and remediated to meet customers’ performance expectations, but the tremendous redundancy makes it hard to identify exactly where the failures originate. In this talk, I will briefly overview the architecture of cloud-scale networks and then describe the automatic network management systems and tools we use in combination with Software Defined Networking to achieve our goal of making Azure the most available network in the world.
Increasing cloud adoption is enabling the next generation of scale-out applications and enabling enterprises to have a diversification strategy as their future-proof model. Customers expect seamless integration across platform and environments, with extended private space, consistent security models, workload mobility, common provisioning and automation tools and open source as de-facto standard.
Google Cloud Platform embraces an open ecosystem in which hybrid and multi-cloud will be the norm. Join us to learn how Google Cloud Platform is enabling customers to deploy multi-cloud and hybrid solutions, to seamlessly extend their existing private and public environment, while opening the potential of Google scale in services, performance and analytics.
Over the past 20 years, there has been little to no innovation. Instead, one sheet metal box is traded for another. But due to the advent of software defined networking, times are changing. Real innovation is happening. How did this come about? What factors drove it into being? This talk will focus on a journey to create a new model for remote branch networking. Key takeaways –
Augmedix frees doctors from the burden of documentation, allowing them to focus on what they do best: taking care of patients. Augmedix empowers doctors to get out from behind the electronic health record to re-humanize doctor-patient interactions. Founded in 2012, Augmedix is the first and largest Google Glass company. Come participate in a conversation with Ian Shakil, the company’s Cofounder and CEO, to discuss:
Like many companies, Thomson Reuters is embracing next generation cloud capabilities. With over 200 Thomson Reuters data center locations and the need to inter operate with multiple cloud providers, this creates unique challenges in network connectivity, security and monitoring. This presentation will walk through the analysis and thought process that Thomson Reuters went through and explore challenges that are still without solutions. The current solutions implemented will be shared, along with the future roadmap.
The relationship between networking and security has been very close for the last 20 years. It wasn't possible to secure the applications so security became a networking problem. Most networks use firewalls, IPS, proxy servers and other middleboxes for traffic interdiction. But times have changed. Once there were many network protocols, today its only HTTP, DNS and SSH. And the rapid move to encrypt everything, including DNS, means that middleboxes less metadata on which to implement security policy
In order to offer baremetal services, operators are either forced into vendor specific solutions or they use flat networks to deal with the provisioning of bare metal servers. While this works in private cloud or smaller deployments, this presents an issue in the public/hybrid cloud or larger deployments, where multi-tenancy is critical. There is no viable open source solution for this. Arista took upon the challenge to address this in OpenStack community by leading and driving Neutron-Ironic integration. Now operators can deploy bare metal servers in a similar manner as they are used to deploying virtual instances. Come and learn the details of this architecture and see how operators are utilizing this and how you can take advantage to build true multi-vendor open source based deployments.
Networking is a critical component of any cloud. When operating a cloud at scale, it's critical to architect and run only the pieces of the stack specific to your deployment. It's also important to empower your developers to be successful. Understanding that you can use DevOps principles for your networking components opens up a whole new world for your developers. Using OpenStack Neutron, Open Virtual Networking (OVN), and Open vSwitch (OVS), IBM is building a next generation public cloud to scale for our clients. We have built a team in-house which is working with upstream Open Source communities to deliver scalable networking for our public cloud. Early on we realized the path to controlling our own destiny and enabling our success was to work upstream. Thus, we've built a system which allows us to track changes from master branches and deploy the latest code from OpenStack Neutron, OVS, and OVN, amongst other components. Utilizing Open Networking technology, we've been able to move swiftly to deliver scalable networking for our public cloud. In this talk, I will cover how we've successfully enabled our developers to move fast and work with upstream projects, the technologies we use to build our networking layer, and where we'd like to go in the future.
Imagine networks in the future. Operators deploy and manage best-of-breed technology from any vendor. These platforms offer a common API that enables programmatic management without the need to support multiple proprietary interfaces. Devices stream monitoring data with high frequency and freshness, even on fully-loaded systems with thousands of ports. Software systems use these data streams to make precise adjustments to traffic flows and policies. Network correctness is based on observations of behavior, rather than by scraping device configurations. In this talk we’ll describe some of the progress being made toward this future, specifically by network operators who have started working together to develop open APIs and tools as part of the OpenConfig working group. We will illustrate the solutions using two real production applications, one that implements dynamic route injection for unified communications systems, and the other an intent-based configuration stack used to manage parts of a Web-scale backbone network.
Verizon is actively pursuing an SDN/NFV strategy to modernize its networks and ultimately lower costs while introducing new capabilities. This talk will first provide a brief overview of Verizon’s initial deployments based on OpenStack. It will then describe future directions and technologies Verizon is evaluating as well as lessons learned. In particular, we will discuss ideas for implementing performance-sensitive VNFs and using SDN/NFV to improve infrastructure security.
SAP HANA is SAP’s flagship in-memory data platform for enterprise applications. It adopts a scale-out shared nothing architecture, which is a capable of supporting very large scale deployments and allows applications to be built that otherwise are not possible on traditional RDBMS. SAP offers HANA in a variety of delivery models including on-premise and public / private clouds Unlike many other cloud environments founded on the massive scale-out of relatively small virtualization compute building blocks, SAPs cloud deployments have a significant footprint of bare metal and scale-up systems. High performance and scalable networking infrastructure are critical elements in the streamlined delivery of HANA landscapes to cloud customers. In order to ensure that our customers can fully benefit from HANA’s power, a variety of unique networking challenges had to be addressed in our cloud design and delivery model. These challenges span the range of network development issues - from maintaining high performance to isolation and security to scalability and network manageability. In this session, we present an overview of the network related requirements that have driven our design choices for delivering large scale HANA deployments in the cloud. Additionally, we will present the lessons learned about designing solutions to address network scalability and manageability challenges to satisfy these requirements.
We will cover some of the key design and operation considerations and lessons learnt from Citi Private Cloud networking with IP fabric, network overlays and IP Storage solution to future thoughts about networking in emerging hybrid cloud deployments with container based consumption model.
An overview of networking technology at Bloomberg LP with thoughts on IP storage, multicast, overlays + underlays, security, evolution of the application.
We operate a large infrastructure to power the world’s first social commerce platform, eBay. Managing a web scale infrastructure to support the range from legacy applications to innovative modern ones, while optimizing for user experience and operational efficiency, is a challenging problem. In this talk , we shall present details of our on-going journey from a traditional operator driven infrastructure to a declarative policy based Application-centric Cloud Native model especially as it pertains to Networks and Functions. We shall cover some of our key choices, actions and plans.
Increasing cloud adoption is enabling the next generation of scale-out applications and enabling enterprises to have a diversification strategy as their future-proof model. Customers expect seamless integration across platform and environments, with extended private space, consistent security models, workload mobility, common provisioning and automation tools and open source as de-facto standard. Google Cloud Platform embraces an open ecosystem in which hybrid and multi-cloud will be the norm. Join us to learn how Google Cloud Platform is enabling customers to deploy multi-cloud and hybrid solutions, to seamlessly extend their existing private and public environment, while opening the potential of Google scale in services, performance and analytics.
DC/OS is a datacenter-scale operating system for running modern apps built from containers and stateful data services. DC/OS abstracts your datacenter into a single computer, pooling distributed workloads and simplifying rollout and operations. This new operating model challenges many assumptions about the network, and new approaches are needed. In this talk, Tobi Knaup will introduce DC/OS, and present the challenges of container networking along with the solutions provided by DC/OS and its community partners.