Thank you to everyone who joined SDxCentral and CENX for the September 16th webinar on lifecycle services orchestration (LSO). This fascinating presentation on full services lifecycle orchestration gave attendees a solid understanding of how LSO, network functions virtualization (NFV), and software-defined networking (SDN) components work together and explore the differentiation of network service orchestration versus LSO for user services. Attendees were also able to experience the incredible value of LSO in unlocking the potential opex/capex benefits of NFV and SDN. After the live event, our presenters took some questions. Read the Q&A below.
What are the typical realized cost savings when a mobile service provider installs an Affirmed Networks’ virtual evolved packet core (vEPC)?
Affirmed Networks: It’s a very interesting kind of computation. As you can imagine, the vEPC is a fairly sophisticated set of virtual network functions (VNFs) and functions, especially when you consider what I had described earlier as part of the Affirmed Mobile Content Cloud, and all its value-added services. As you add more and more functions to the core EPC gateway functions, the savings definitely increment quite a bit.
Last year and into the first half of this year, we’ve been generally deployed in quite a few Tier 1 networks, and we did a running study of the migration as those operators moved from legacy-box based functions to a virtualized network with Affirmed Networks’ VNFs. What we noticed was, in a typical scenario when you’re virtualizing the gateway functions along with a few very widely deployed services, the savings come out significantly. On the capex side we noticed savings of around 69 percent over legacy deployments and on the opex side, which was very interesting for mobile operators, the savings were around 67 percent. We did a study with some real world data and use cases, from which we derived those numbers.
Do operators want to manage the NFV infrastructure themselves, or are there operators who want the vendor to manage the NFV infrastructure?
VMware: This is truly up to the operator. They may have the resources to drive and manage their private telco cloud themselves, or they may want a prime SI to operate it for them, or they can choose to buy NFV infrastructure as a cloud service. We are supporting either form factor, providing flexibility and choice.
What is the difference between an NFV infrastructure platform (for NFV workloads) and an IaaS platform (for IT workloads). Are there differences in terms of availability and performance?
VMware: IT is the only way you virtualize the IT workloads, to achieve virtualization. Now we are moving into storage network virtualization as well and it’s not that different from what you would do in NFV, when you talk about the NFV infrastructure layer. It is more about how you design, to make sure that it maintains the application or VNFs on top, so they actually have the ability to fulfill the SLAs required. It’s how you design the virtual infrastructure as a resource cluster.
The other key thing is secured dependencies; we believe that a common platform, a multi-vendor NFV infrastructure platform, is the only way the carriers will truly make the return on investment (ROI) work for NFV. Secured dependencies is the way you’re allocating resources for a container type of approach: You have one container for one VNF, and another container for a different VNF, and you need to have the right resource allocation for compute, network and storage. I believe that is one of the key things to think about when it comes to NFV infrastructure versus IT workloads.
Please explain NFVO-SO versus NFVO-DO again.
CENX: Broadly speaking, the NFVO-DO (Domain Orchestrator), understands the underlying compute infrastructure in terms of the resources available: how much compute is available, as well as how much storage, and how much memory and network are available. It understands how those resources are allocated. For each VNF that it’s managing, it understands how those VNFs are mapped to that underlying compute infrastructure, so it really understands the VNF and how it’s implemented in all that detail of the underlying compute infrastructure.
The NVFO-SO (Service Orchestrator) is sitting on top of that, and what it understands is the VNF in the context of other VNFs, as well as physical network functions (PNFs), in the context of LSO. The NFVO-SO understands the services, the cost, and the combinations of the physical network and VNF that make up those end-to-end services.
It is the build-up of layers, with NFVO-DO essentially providing abstraction of your VNFs and their connection points, and the LSO on top of that understands how those VNFs fit in into the overall network topology and how services are carried across the virtual and psychical infrastructure.
What is service orchestration (SO) and device orchestration (DO)?
CENX: SO is service orchestration, and DO is domain orchestration. Refer to my blog post for more information. Note also that the discussion at ETSI has very recently introduced the notion of a Resource Orchestrator (RO) and the functionality of the RO versus DO is still quite fluid.
Does the LSO concept mesh with Metro Ethernet Forum’s (MEF) definition of LSO?
CENX: Definitely, LSO is being defined by MEF, and CENX is actively participating in the definition of its capabilities and APIs. There are efforts to define the LSO Reference Architecture in particular.
How quickly can a service provider integrate and deploy CENX’s LSO with NFV components and existing OSS? What APIs are used?
CENX: The timeline to deploy the CENX’s Cortx Service Orchestrator really depends on the number and complexity of the integration points. A typical Tier 1 carrier integration, involving multiple NE and VNF types and four to five OSS, would take about six months. APIs vary depending on the VNFs, NEs, and OSSs below the LSO software, e.g. CORBA, REST.
For what service provider use cases has VMware’s NFVI been deployed?
VMware: VMware has been deployed in use cases such as vEPC, VoLTE, IMS, Wireline vCPE, to mention but a few. As mentioned in the presentation, we distinct ourselves in two solution areas, vEMS (Virtual Elastic Mobile Services) and vWMS (Virtual Wireline Managed Services).
How many subscribers currently get their services from Affirmed infrastructure?
Affirmed Networks: Affirmed Mobile Content Cloud today provides services to millions of subscribers including those in Tier 1 mobile networks. Beyond this level of maturity of our solutions providing services to a large number of subscribers, what makes us at Affirmed really excited are also the facts that these subscribers span across multiple tier networks — Tier 1, Tier 2, Tier 3 — given the great economic elasticity that NFV now affords across all Tier operators. The vast breadth of application and use cases that our solutions enable for these subscribers are:
- Full suite of smartphone consumer traffic
- Smart device (phone, laptop, server, enterprise device) enterprise traffic
- IoT and Machine to Machine traffic across multiple business and industry verticals
- Multiple Tier MVNO’s ranging from those that provide specific services for niche markets, to those MVNOs that provide a broad set of services and own their own part of the infrastructure.
This level of maturity and flexibility from a truly virtualized solution that enables multiple business models is the key to Affirmed Networks’ solutions being deployed across the globe at multiple Tier mobile operators.
Does the LSO implement multi-tenancy? Is it exposed to end customers to add services, or just used by the operator?
CENX: Implementation of CENX’s Cortx Service Orchestrator varies depending on the operator and service provider. In some of our implementations, end customers indeed have direct access to the system via a customer portal. In this case, of course, they can see change and add only their own services and are limited in their view of their services (they see only an abstract view of their services across the service provider cloud, not the details of how they are implemented inside the cloud). An operator logged into the system has broad access across customers and can see the details inside. Indeed, it is critical that operations group within the service provider knows more than their customers. In other implementations, the service provider does not expose any direct interfaces to customers.