John Hawkins Ciena Senior Advisor for Product and Technical Marketing (AKA @EtherJohn), received so many questions after a webinar on the benefits of adding packet services to an OTN networks, that we thought we answer them here
Q: How does the new world of software defined networks (SDN) and network functions virtualization (NFV) factor in to packet and optical transport network (OTN) services? Does one rely on the other, or will SDN make both OTN and packet services obsolete?
JH: This is a really involved question, but the short answer is that SDN and NFV services are dependent on reliable connectivity, both to deploy the virtual function and to interconnect these services. SDN forwarding planes can be used to create that connectivity, which could be a point-to-point e-line service, or a multipoint to multipoint e-LAN service. NFV, in particular, will allow service providers to better differentiate as connectivity becomes commoditized by allowing them to layer on other offerings such as a firewall service, SD-WAN service, virus detection and deep packet inspection, to make them more valuable to their end customers. All that to say SDN and NFV are interrelated with OTN, so one doesn’t make the other obsolete.
Q: How can service providers avoid cannibalizing my high-value OTN services with Ethernet or packet services; don’t they compete?
JH: There are a variety of use cases for each technology. For example, in some cases OTN services would be overkill – and too expensive – for a specific customer’s requirements. Ethernet services, even with protection and high service level agreement (SLA) boundaries built in, are often more cost effective than equivalent OTN services. This is due to factors such as statistical multiplexing, the granularity of the services, and their dynamic nature that makes them easy to spin up or down very quickly. So while there is overlap, they don’t really compete and providing customers a choice means you can offer them the right service at the right price, even mixing the two and offering them as a hybrid service, given the right architecture.
Q: Aren’t packets and Ethernet only appropriate for best-effort services, is that no longer the case?
JH: Definitely not! While Ethernet can be used when best-effort is required, such as over-subscribed, unprotected services that don’t have a committed information rate, but this isn’t the only use case. Carrier Ethernet allows network operators to guarantee a service and provide an SLA, all while minimizing the effects of Spanning Tree Protocol that may have limited the potential applications of Ethernet in the past.
Q: It appears multi-layer integration is a good thing, but doesn’t SDN manage each layer (layer 0, layer 1 and layer 2) separately?
JH: Yes and no. Service providers may want to manage factors such as protection over multiple layers individually. Most of today’s SDN technology operates at the packet layer, at the switch and IP level. However, the standards bodies that manage OpenFlow and other SDN approaches are also considering layer 0 constructs, such as path selection, creating bandwidth, expanding bandwidth, and bringing down services, all using centralized SDN control planes. All that to say the technology is evolving into a multi-layer SDN capability over time, even though it’s early days for SDN as a carrier technology, so it’s not yet widely deployed.
Q: Can service providers apply a bandwidth profile on a per-interface basis for the packet interface in SDN?
JH: Yes, service providers can apply a bandwidth profile on an interface basis. Alternatively, bandwidth profiles can be applied on a service basis using a VLAN, for example, or using a virtual switch.
The webinar, There’s Gold in Your OTN Nework, Mine it with Packets is available on-demand HERE.