When we talk with communications service providers (CSPs) about our SD-Core solution, one of the most common questions we get is, “Is this like SD-WAN?” The short answer is no, not at all, totally different. They are so different, we weren’t sure we should even write this clarification for it’s like comparing apples to oranges. In fact, it would probably be quicker to list the things that SD-Core and SD-WAN have in common. They both start with “SD,” and… that’s about it.
Of course, that common acronym for “software-defined” likely contributes to the confusion. So, let’s break down the differences.
Different Solutions for Different Users
For CSPs, SD-WAN is typically a third-party connectivity solution that you resell to business customers to connect their branches. The “SD” part refers to how this solution decides which network path to use for different applications, usually among public broadband connections, for which SD-WAN acts as a logical overlay – not necessarily centralized SD controls, per se. Note that it’s the third-party SD-WAN solution making these decisions, not the service provider.
SD-Core, on the other hand, is a tool that CSPs use to optimize internal traffic management over their own infrastructures. With SD-Core, you can define the specific network paths you want traffic to use when traversing your network, and automatically establish and monitor those paths.
For example, you can map individual enterprise customers to their own paths one-to-one. Or, you can aggregate traffic from multiple customers along the same path, based on whatever type of classification you implement to support your business plans. In either case, you can classify traffic in a way that’s much simpler and more scalable than traditional, distributed traffic engineering approaches like RSVP-TE or LDP.
Different Use Cases
Another big area of differentiation is use cases. For SD-WAN, there’s just one: connecting an enterprise’s branches to a central data center and/or each other. SD-Core, on the other hand, can enable a wide range of use cases, including:
- Core Offload: Use SD-Core to extend your current CapEx investments by decongesting the expensive MPLS core and diverting less critical traffic to a parallel core built from “white box” network elements. Telstra is using this model right now to support advanced analytics. They can uncover deep insights by routing random samples of network traffic to a new analytics server farm—without having to consume revenue-generating core network capacity to do it.
- Core Optimization: Optimize the existing MPLS core by using centralized traffic engineering (Segment Routing), which scales much better than conventional distributed models. Many service providers find this extremely useful in outfitting their existing “brownfield” core networks to support new capabilities and software-defined automation.
- Network Slicing: Map traffic from different applications to “slices” of the network that support specific requirements or 5G use cases. For example, automatically route business video traffic over one slice optimized for enhanced mobile broadband (eMBB), while directing customer IoT traffic towards the slice designed for Massive Machine-Type Communications (mMTC).
Another way to think about the difference between the two solutions: SD-WAN is one type of logical overlay, typically, that third-party vendors (like VeloCoud, Cisco Viptela) can operate on top of your physical network infrastructure. SD-Core is a tool you can use to build your own logical overlays for your infrastructure, as well as perform other types of traffic engineering. You could use SD-Core to create an SD-WAN offering. But you could also use it to do much more.
|SD-Core vs. SD-WAN Cheat Sheet|
|Who is the user?||Enterprise customers||Service providers|
|What does it do?||Connects distributed offices with a central data center and each other.||Helps service providers customize traffic-handling over their networks—whether traffic comes from enterprises or the CSP’s own edge traffic.|
|What kind of software-defined intelligence does it provide?||Can identify different types of application traffic (voice, video, web conferencing) and route each along a different logical path.||Provides fine-grained traffic engineering capability to define logical paths across the network for any purpose (e.g., for specific 5G use cases, customers, SLA tiers), and automates setup and monitoring of those paths in the network.|
|Who controls the SD traffic handling?||The SD WAN solution vendor. The CSP has no control over how traffic is routed.||The service provider. Telcos use Lumina SD-Core APIs to directly control traffic grooming on the network. Typically, they use standards-based mechanisms to define traffic paths such as BGP, MPLS-TE, or Segment Routing.|
|Where does it operate in the network?||Typically functions as a logical “overlay.”||Optimizes traffic in the network “underlay.”|
|Which use cases does it support?||Branch connectivity||SD-Core enables a wide range of use cases, including:
|What standard does it support?||SD-WAN is a proprietary solution and locked-in to a specific SD-WAN vendor. The Customer Premise Devices are proprietary, and so is the interface between the CPE and the cloud-based SDWAN controller. There is no mix-and-match possible owing to a lack of standardization.||Luminas SD-Core is a standards-based offering and it supports the IETF Segment Routing RFCs. It uses the standards prescribed for centralized traffic engineering which touches upon standards across BGP, PCEP and many IGPs. For the OpenFlow and optical domains, it uses standards prescribed by communities from those domains.|
If you’re imagining the possibilities that scalable traffic engineering and customized network paths can bring to your organization, it’s easy to get started. Lumina offers the SD-Core solution for three types of networks: optical, packet, and white-box/flow-based networks that use OpenFlow.
In each case, the solution includes:
- Lumina SDN Controller, our quality-assured edition of the industry-leading OpenDaylight open SDN controller
- Lumina Extension & Adaptation Platform (LEAP), a model-driven software platform to automate heterogeneous networks
- Lumina’s Path Computation Engine, a domain-agnostic path computation engine that can work for a wide variety of networks such as Ethernet, IP/MPLS, Optical, OpenFlow, P4 etc. It supports a wide variety of constraints as well
- Lumina Network Resource Manager (NRM) Applications, which provision and monitor traffic-engineered paths across the different types of networks—packet, optical, and flow.