Thanks to everyone who was able to join us on August 23 for DemoFriday™ with Cyan and Accedian Networks. If you missed the demonstration on multi-layer, multi-vendor WAN provisioning and cloud service orchestration, it now is available in our archives.
The presentation demonstrates a real-world software-defined networking (SDN) use case that enables end-to-end orchestration of cloud services from the enterprise to the data center, across the WAN. Watch the full presentation, download the podcast, or check out the teaser video and other resources below.
During this exceptional demo, participants asked some key questions, and we’ve provided them for you here, with responses from Nirav Modi of Cyan and Ken Gold from Accedian. Read on:
Q: I understand that SDN is not OpenFlow. But I am curious on which part of this demo collecting stats or creating flows was done using OpenFlow? I saw from earlier slide that the Ryu controller is used. Hence, I guess OpenFlow is used.
Nirav Modi, Cyan: The service-orchestration application that was demonstrated used a Ryu OpenStack Neutron plug-in (formerly called Quantum) to control the Open vSwitches inside the datacenter. This Ryu plug-in makes use of the Ryu OpenFlow controller, which communicates OpenFlow to the Open vSwitches.
Q: In the earlier deck presented by Cyan, I understand that Blue Planet is controlling the enterprise, data center and carrier networks. I also see that Blue Planet is calling the OpenStack API. Does Blue Planet expect OpenStack to be deployed in enterprise/data-center networks?
Nirav Modi, Cyan: The end-to-end service orchestration provided by Blue Planet supports OpenStack-enabled data-center networks today. However, the architecture is open and could be extended to support other virtualization platforms, such as VMware.
Q: Can you change the CIR without stopping the file transfer?
Nirav Modi, Cyan: Yes. Within the WAN bandwidth calendaring feature that was demonstrated, the bandwidth profile can be modified while applications are running, e.g. the backup application. The changes are applied immediately and do not disrupt or terminate the sessions that are in progress.
Ken Gold, Accedian: Yes. Changing bandwidth profiles without having to delete and restore the service is something we have supported all along. Having this capability allows SDN controllers, such as Blue Planet, to react in real time to changing network and service conditions.
Q: “Multi-layer” and “multi-vendor” are great, but what about multi-network, or many service providers along the end-to-end circuit?
Nirav Modi, Cyan: The concept of multi-domain orchestration is core to the Blue Planet architecture, where a domain could represent a different service provider network, a different data-center provider, or a different affiliate within a single service provider.
Q: Does this work with virtual machines from a public cloud provider like Amazon Web Services (AWS)?
Nirav Modi, Cyan: Cyan’s Blue Planet integrates with OpenStack-enabled data center networks today, but the architecture is open and can be extended to orchestrate to a public cloud such as AWS.
Q: Can we implement bandwidth changes without breaking file transfer?
Nirav Modi, Cyan: Yes. As mentioned previously, the WAN bandwidth calendaring feature that was demonstrated supports real-time changes to the bandwidth profile while applications are running.
Ken Gold, Accedian: Yes. Supporting this feature in the network elements is what allows SDN controllers, such as Blue Planet, to dynamically adjust network resources to adapt to changing network and service conditions.
Q: How do the concepts of oversubscription equate to SDN?
Nirav Modi, Cyan: Having a centralized control plane, which is a core concept in SDN and Cyan’s Blue Planet SDN platform, provides the network operator with full visibility of network resources, including bandwidth. With this centralized resource visibility, the network operator has greater control of network oversubscription and could dynamically change oversubscription based on policy.
Ken Gold, Accedian: Oversubscription is a fundamental piece of today’s packet network and is what helps to drive network efficiency. Being able to visualize, in real-time, KPIs such as bandwidth utilization, latency, and packet loss with a rich API like Network State Plus is what allows SDN controllers, such as Blue Planet, to dynamically adjust to the ever-changing conditions of an oversubscribed network.
Q: For enterprise customers, what are some of the use cases that you are seeing them implement this for?
Nirav Modi, Cyan: We are seeing a trend where large enterprise customers are demanding more control over their end-to-end user-experience. Many network/data center operators are not able to offer a unified service, and their enterprise customers (end-users) often experience poor application performance. To circumvent this, some large enterprises are choosing to build their own private networks and private clouds – so they can holistically manage the entire user experience. Applications range across the board, from PaaS, IaaS, and SaaS to hosting, streaming and CDNs.
Q: How are SDN and OpenFlow different from an EMS? The demo specifically touched on the EMS functions, but how is all this translated to SDN/OpenFlow?
Nirav Modi, Cyan: As showcased in the demo, a carrier network is often a multi-vendor environment. To this end, we orchestrate the services using various interfaces towards the WAN network elements, including SNMP and CLI as well as indirect provisioning through an existing EMS or NMS. In the data center, where OpenFlow-enabled devices are more readily available and gaining traction, we use OpenStack APIs to manage the virtual and bare-metal resources, as well as Neutron plug-ins to drive network configuration. Some of the plug-ins (e.g the Ryu plug-in) use OpenFlow to configure the data-center Open vSwitches.