Loading…
OCP US Summit 2016 has ended

Sign up or log in to bookmark your favorites and sync them to your phone or calendar.

Networking [clear filter]
Thursday, March 10
 

9:00am

Open Networking Moves from Data Center Top-of-Rack to Network Core and Network Edge
View live stream at: http://www.youtube.com/watch?v=G5f3AW_avdI


Open Networking in Open Compute has grown from simple hardware only designs for data center Top or Rack (ToR) switching to a complete ecosystem of hardware providers, commercial network operating system providers, open source applications, and open source network operating systems. The benefits realized by Open Networking is much more than a simple reduction in capex spending it is about reduction of vendor lock in with freedom of choice, rapid innovation through fast paced start-ups and community based projects , and expedited delivery of new services to end users. Open Networking is now migrating to the center of network topologies while at the same time the access layer of networks. The open networking benefits realized by data center operators is now being pursued by Telecomm operators and also those providing access layer services to Campus and Enterprise customers. Accton and Edgecore will presenting the next chapter in Open Networking with contributions to OCP including a large scale open networking modular system. This modular open networking system is envisioned as the next generation open architecture platform that will allow for the integration of various networking technologies in a common cost effective open form factor. 

Accton and Edgecore will also be presenting contributions to Open Compute that allow open networking to now reach the access layer of networks though contributions of various access layer devices.  It is envisioned that these access contributions will bring the same open networking benefits realized by data center operators, along with the establishment of a new ecosystem of software solution providers, to the Campus and Enterprise markets.

Speakers
JC

Jeff Catlin

Associate Vice President of Technology, Accton Technology Corporation


Thursday March 10, 2016 9:00am - 9:30am
Lower Level: LL21A/B

9:30am

Ensuring Interoperability and Open APIs for Open Networking Systems
Live Stream - http://www.youtube.com/watch?v=1vGxK2gB2zo

Open Networking has created new opportunities and new challenges for Data Center designers and implementers. Many customers are used to using monolithic solutions from a single integrator. When using Open Networking solutions from multiple providers, a few key questions arise: will it all work? Are there common interfaces I can use to monitoring and control the hardware? How do Data Center designers and implementers ensure that the components they’ve chosen will be interoperable?

This presentation will cover two key projects related to Interoperability, underway in the Networking group:

1. Interoperability Testing: the content will cover
a) a brief history of the project at the University of New Hampshire InterOperability Lab to test and validate combinations of OCP-certified switches, network operating systems and interface modules,
b) outline the test procedures for ensuring basic interoperability of Open Networking components, including proper physical layer operation for cables, pluggable modules, and switches, as well as baseline packet loss expectations, and
c) plans for further expansion of the testing in 2016.

2. Open Optical Monitoring: the content will cover
a) a brief description of the need for a common set of APIs to access information and control pluggable modules deployed in Open Networking solutions,
b) a review of the project to date,
c) a demonstration of available code enabling ‘Open Optical Monitoring’ and d) plans for the project in 2016.

Speakers
avatar for Craig Thompson

Craig Thompson

Director of Strategic Marketing, Finisar
Craig Thompson is currently Director of Strategic Marketing at Finisar Corporation. In his role Craig focuses on long-term product and technology roadmaps for optical communications, and contributes to advanced development activities supporting the roadmaps. Craig has held a number... Read More →
avatar for David Woolf

David Woolf

University of New Hampshire InterOperability Laboratory (UNH-IOL)
David Woolf currently leads Data Center initiatives at the University of New Hampshire InterOperability Laboratory (UNH-IOL). While at UNH-IOL, David has led development of dozens of industry-reviewed test programs and procedures as part of the team that has grown the UNH-IOL into... Read More →


Thursday March 10, 2016 9:30am - 10:00am
Lower Level: LL21A/B

10:00am

Switch Abstraction Interface (SAI)
Live Stream - http://www.youtube.com/watch?v=UPuqCX4PLmw

Switch Abstraction Interface (SAI) has been officially accepted by the Open Compute Project (OCP) as a standardized C interface to program hardware ASICs. SAI came into existence as network operators started seeking ways to adopt a more dynamic, programmable infrastructure. The lack of a common and well understood way to program network switch ASIC’s is a source of reduced agility, increased complexity, poor reliability, and greater cost. SAI helps by being a standardized C API to program ASICs.

SAI allows software to program many different switch chips without undergoing any changes. This helps us to keep the base router platform simple, consistent, and stable. It also reduces the time to market, and enables the adoption of the latest available hardware easily. It breaks the software-hardware coupling and allows us to pick the best fit of software and hardware on a “need by application” or a “need by network” basis. By providing simple, consistent interfaces for applications and protocol stacks that orchestrate and automate cloud services, it helps consume the underlying complex and heterogeneous hardware easily.

This helps in shifting our focus to applications that require integrating our network with our cloud. SAI is, therefore, a big step towards open networking software. Today SAI has 10 contributors, and features that range from basic L3 functionality to more complicated QoS, PFC with RDMA, and buffer management. The SAI demo hosted live at the Microsoft booth will display these features and the power of open networking by integrating 5 different NOS, 5 different switch platforms, 5 different ASICs, over 6 *same* SAI implementations.

Speakers
avatar for Kamala Subramaniam

Kamala Subramaniam

Architect, Microsoft
Kamala Subramaniam is an architect in Azure Networking at Microsoft. She drives the success of the Azure Cloud Switch and the adoption of the Switch Abstraction Interface. She works closely with several operators and vendors in the industry at the OCP to innovate the open networking... Read More →


Thursday March 10, 2016 10:00am - 10:30am
Lower Level: LL21A/B

10:30am

Azure Cloud Switch (ACS)
Azure Cloud Switch (ACS) - Live Stream - http://www.youtube.com/watch?v=5XFm9EteAks

The cloud depends on high-speed, highly available networks to power their services. This makes it critical for network operators to be able to control their own destiny by rapidly adding to their network features they need while keeping out feature changes that increase risk and complexity. ACS came into existence to simplify the way we manage our data center networks. Part of that challenge was to manage the fleet of switches like we manage our servers. To be able to use new switching ASICs from various switch ASIC vendors to reduce time to market. To be able to develop new applications on the switch easily. ACS uses Linux as the base operating system and breaks the software running on the box into different modules. A set of software modules are developed to manage the switch ASICs as well as software routing stack such as BGP. Like any other software modules in Linux, users can use common Linux tools to manage these modules independently.

For example, user can use popular Linux automation tools such as Puppet, Chef and Ansible to manage them. This makes managing this Linux switch box no different than managing all the servers in our data centers in principle. As a result, we integrate this Linux-based box with our network management stack to enable some agile software development practices such as weekly software upgrade and quick bug fix rollout. Azure cloud switch software stack runs on top of Switch Abstraction Interface (SAI), which allows the same software stack to run across different switch platforms. SAI provides a set of unified switch control C APIs across multiple ASICs. Azure cloud switch provides a shared storage for all switch states, which allows users to easily develop new applications to manage the switch states. We developed a switch state service which manages all states within the switch. It provides a consistent view of all switch states to any application that subscribes to them. Above the switch state service, developers can then develop different applications to manage any switch based on the up-to-date view. Below the switch state service, sync agents synchronize the switch states to the hardware throughout SAI.

In summary, this demo presents ACS as a collection of software modules with Linux as the base OS, that runs on top of several switch ASICs, Broadcom, Mellanox, Cavium, Centec, Mediatek, and Barefoot. It allows user to easily integrate switch with existing automation and management software to support agile software development and deployment. The software stack also provides a central place to store all switch states and allow multiple applications orchestrate the switch resource management.

Speakers
avatar for Kamala Subramaniam

Kamala Subramaniam

Architect, Microsoft
Kamala Subramaniam is an architect in Azure Networking at Microsoft. She drives the success of the Azure Cloud Switch and the adoption of the Switch Abstraction Interface. She works closely with several operators and vendors in the industry at the OCP to innovate the open networking... Read More →


Thursday March 10, 2016 10:30am - 11:00am
Lower Level: LL21A/B

11:30am

Update on OpenSwitch with OCP Projects
Live Stream http://www.youtube.com/watch?v=sgWCPxt4tCI

In this talk we will show the current status from the OpenSwitch NOS (www.openswitch.net) running on the different OCP networking hardware, and updates on the roadmap and future plans of the project.

Speakers
avatar for Diego Dompe

Diego Dompe

Senior Software Architech, Hewlett Packard Enterprise
Diego Dompe is a Senior Software Architect in Hewlett Packard Enterprise Networking, and an Open Source and Linux expert. He is an specialist on software-hardware interaction and switching silicon.


Thursday March 10, 2016 11:30am - 12:00pm
Lower Level: LL21A/B

12:30pm

How app-enabled switches increase innovation for telcos and DC operators

http://www.youtube.com/watch?v=aRnh_e3wpDc

Telco and data centre operators want software-defined capabilities to match their newly open hardware. This talk will demonstrate the state of the art for top-of-rack switches, as well as forthcoming telco-centric open networking such as DSLAMs, GPONs and software-defined LTE base stations, that build a software ecosystem to take advantage of OCP’s open hardware strategy.

Using open source mechanisms optimised for scale-out environments, the speakers will walk through a model central office or data centre, showing how diversity of NOS options and diversity of networking apps can fit together in a highly-automated cluster. Using OpenBMC and FBOSS from Facebook, MAAS, LXD containers, Ubuntu and a range of apps, the speakers will demonstrate dramatically more efficient software-defined approach to infrastructure.

We will use OCP switches to demonstrate running apps on the ToR switch that deploy and manage both server and physical switch infrastructure. Using OpenBMC, PXE, and ONIE, we are able to automate the deployment of scale-out bare metal infrastructure using 100% open source tools.  We will show how the separation of OS and apps enables greater competition for network control plane vendors as well as creating new opportunities for innovation in the SDN, NFV and network intelligence markets.


 


Speakers
avatar for Mark Shuttleworth

Mark Shuttleworth

CEO, Canonical
Mark leads Canonical products and is founder of Ubuntu, the reference operating system for OpenStack clouds. Mark has charted the course that made Ubuntu the top OS for all public and private cloud platforms, and brings his passion for next-generation platforms and technology to this... Read More →
avatar for Luke Williams

Luke Williams

Technical Partner Manager, Canonical
Luke Williams is a systems network engineer. He has worked professionally in IT for over 22 years in different roles and has over 10 years of experience in networking architecture and engineering. He has multiple certifications from Cisco, Silver Peak, Juniper as well as Red Hat... Read More →


Thursday March 10, 2016 12:30pm - 1:00pm
Lower Level: LL21A/B

1:00pm

Linux Networking is Awesome
Live Stream - http://www.youtube.com/watch?v=N-8aAWqh004

The Linux kernel has seen some significant improvements in networking over the last several years that make it attractive for building networks with it. Improvements such as: scaling interfaces, vlan aware bridges, VRF, MPLS, and switchdev empower almost all Linux distributions to drive network switches at line rate. When combined with OCP projects like ONIE and APD and OCP hardware like the AS5712 or the MLX1410, one can see how powerful the base Linux kernel has become. We plan on demonstrating this power live.

Speakers
avatar for Scott Emery

Scott Emery

Principal Engineer, Cumulus Networks
I'm the OCP Networking Project Co-Lead and an Engineer at Cumulus Networks. I'm interested in pretty much anything related to networking, and especially open networking. Or, the San Jose Sharks.
avatar for Shrijeet Mukherjee

Shrijeet Mukherjee

VP of Software Engineering, Cumulus Networks
Shrijeet is the VP of Software Engineering at Cumulus Networks. Most recently he was Director of Engineering with the team that delivered the highly successful VIC adaptor line in Cisco's UCS servers. His career has spanned a wide variety of technologies including Team Lead of high... Read More →


Thursday March 10, 2016 1:00pm - 1:30pm
Lower Level: LL21A/B

1:30pm

Facebook Wedge 100/6 Pack hardware Deep Dive & Hardware Testing
Live Stream - http://www.youtube.com/watch?v=AI0j_4CvrEw

Hardware design deep dive of following FB network projects
1: wedge100 32x100G TOR switch
2: SixPack 128x40G Modular Switch

Speakers
avatar for Hany Morsy

Hany Morsy

Network Engineer, Facebook
Hany Morsy is a network engineer at Facebook on the data center network engineering team. Currently his focus is on the large-scale testing infrastructure for Facebook’s data center switches and topologies.
avatar for Zhiping Yao

Zhiping Yao

Network hardware engineer, Facebook
Zhiping Yao is a network hardware engineer at Facebook, working on Facebook networking hardware such as the Wedge and 6-pack data center switches. Currently his focus is on Wedge 100. Prior to Facebook, Zhiping worked at Brocade on high-speed data center and campus switches.


Thursday March 10, 2016 1:30pm - 2:00pm
Lower Level: LL21A/B

2:00pm

How does FB manage thousands of OCP switches?
Live Stream - http://www.youtube.com/watch?v=fMeECyHaxYw

Facebook has deployed thousands of OCP-accepted Wedge switches into its datacenters over the last year and that number continues to grow. The FBOSS software evolved rapidly throughout that period as we dealt with large-scale outages, different topologies, and other software challenges. At this point, we are able to update the software on thousands of our switches at least weekly, often without dropping a packet. Further, the Facebook 6-pack switch, which we have shared with the OCP community in advance of the OCP summit, is now in production as well. 6-pack is a 128x40 Gb switch designed to operate as the fabric/spine of our datacenter fabric, and it presented several major challenges in the control plane as well as in the management plane for the FBOSS and network management team. In this workshop, we will share our development and operational best practices with the OCP community.

Our FBOSS and network management software engineers will cover how we addressed the challenges of getting FBOSS and our attending operational software/processes to handle thousands of Wedges and how we expanded the same FBOSS and operational software to handle 6-pack. They will cover the network design philosophy of simplicity and scalability and how the hardware and software are crafted to support that.

Speakers
avatar for Jasmeet Bagga

Jasmeet Bagga

Software Engineer, Facebook
Jasmeet Bagga is a software engineer at Facebook, working primarily on network infrastructure projects.  Currently his focus is on FBOSS, Facebook's switching and routing software stack. Prior to Facebook, Jasmeet worked at a start-up building products for network diagnostics and... Read More →


Thursday March 10, 2016 2:00pm - 2:30pm
Lower Level: LL21A/B

2:30pm

Enabling Pervasive Network Security Using OCP Switch Hardware and ONL
In this talk, the speaker(s) will present a new architecture for enabling pervasive network security and monitoring capability in a large-scale data center or campus network. Key Architectural Requirements:
* Monitor E-W network traffic in addition to N-W traffic (variable granularity and physical+virtual workloads)
* Optimal utilization of security, monitoring, performance tracking tools
* Scale-out Design
* Economical at Scale In addition to sharing the architectural requirements, speaker(s) will describe an actual deployment of the solution in a large financial services data center. Salient features of the solution include:
* 30+ OCP switch based architecture in a multi-tier design * Open Network Linux (ONL) based switch software -- specialized for working in a controller based architecture
* OCP/ONL-based Switch Economics (compared to a traditional closed, proprietary solution)

Finally, the talk will discuss the applicability of this new, OCP centric design to other large-scale deployments in service provider as well as enterprise environments.

Speakers

Thursday March 10, 2016 2:30pm - 3:00pm
Lower Level: LL21A/B