The glistening blue waters of the Mediterranean Sea served as a splendid backdrop to the TM Forum this year, where a significant amount of time was dedicated to the management of SDN and NFV. Indeed, there was even a specific conference track that spanned the first two days of the show. Clearly, there is a lot of hype and discussion surrounding NFV, but where is the market? Where is the money being spent?
If the transition to virtualized environments in the enterprise is any indicator, it appears that the transition to NFV in CSPs is only a question of when, not if. Why is this so? Well, for the same reason the enterprises did it: to be more agile. Sure, there is the argument to be made for reducing the number of physical servers, saving money, etc. But if we have learned anything from our enterprise counterparts, the real benefit is agility. Why is this important for CSPs? The reasons include:
- Video: The main reason mobile users continuously ask for more bandwidth is to run video applications, which can be satisfied only through constant upgrades of infrastructure capacity.
- Internet of Things (IoT): The massive adoption of IoT devices will put additional strain on already overloaded networks due to huge numbers of new “users” (referring to sensors) that require high service availability and compete for limited network resources with actual human users.
- Self-Optimizing Networks (SON): In order to respond to rapidly changing requirements (think congestion around a football match or daily fluctuations in traffic), CSPs will require dynamic environments that can be adjusted based on traffic automatically and without rolling trucks. This points to the importance of orchestration, not to mention RTTMS!
So, it’s clear that trying to upgrade the existing “physical” network will not be sustainable from CAPEX, OPEX and flexibility standpoints. As CSPs face a future with potentially decreasing ARPU, the ability to quickly respond to customer needs of may be the difference between thriving and surviving.
Furthermore, operators must also change their approaches regarding how they offer services: The days of flat ratesare over. It is no longer possible to attract customers with an “all-you-can-eat”offering because the risk is high that service quality will decrease until it is under the minimum acceptable level. Perhaps such an approach would still be good for pre-paid or value-based contracts offered to young users, but not certainly for high-value accounts like business users who are willing to pay more, but only for a guaranteed level of service. Since it’s not possible to please everyone, CSPs must differentiate their service offerings through increased flexibility and agility. Today, typically all service plans provide a fixed set of features over a pre-defined time range (measured in years); tomorrow CSPs must be able to offer services to match user demands, however briefly that service is required. Most CSPs have already realized that a transition to NFV will not happen all at once; it will be a “journey” and they must be very careful in adopting a new and disruptive technology. Again, why is that? It’s because NFV technology will reshape not only the network, but also the organization (and, more importantly, the culture) behind it. TM Forum presented a good opportunity to understand where organizations are on this journey.
Virtually (no pun intended) all of the operator solutions presented about real NFV implementations showed examples delivery of “Enterprise Connectivity Services” (a.k.a., Software-defined Wide Area Network, or SD-WAN) through vCPE (Virtual Customer Premises Equipment). The concept is to deploy COTS hardware at the customer premises (initially designed for the enterprise market, but this is also extensible to residential implementations) instead of dedicated pieces of hardware (e.g., IDS, Firewall, Access Routers, etc.) whose functions are then defined by software and applications that are configured remotely with as little manual intervention as possible (actually, the target is the so-called “zero-touch”or full automation). All of these solutions, already in or very close to production, are based on providing the end -user a Web portal where he/she can select the services to enable (e.g., increase bandwidth on the connections between branch offices). The network can then automatically provision the required services programming the COTS hardware previously installed at the customer premises. This means that a minimum level of physical installation is still required for the on-site vCPE at startup, but then the operator has great flexibility in service provisioning; rather than sending installation personnel to customer locations, they can allow the end user to activate the services they need. This translates to many benefits, such as:
- Reduced costs for CAPEX, due to the adoption of standard COTS hardware instead of multiple dedicated appliances, and OPEX, thanks to easier maintenance.
- Rapid delivery of new services (operators cited less than 1 minute, compared to many days as of today).
- Increased service flexibility (for example, an enterprise could pay just for a single videoconference and not for the bandwidth to support such service annually).
Another NFV use case, but not as developed as vCPE from a full production standpoint, is the vEPC (i.e., mobile core network based on virtualized nodes). In this area, the approach is to deploy a new virtualized infrastructure running in parallel with the legacy one in order to offer a specific new service like VoLTE. In other words, the new service is introduced through NFV instead of upgrading the existing infrastructure; that is still maintained (at least, initially) to provide legacy services.
Clearly, CSPs have realized that they need NFV to better change network capacity dynamically based on user requirements and then enable new business models and revenue opportunities.
However, there are still barriers to NFV adoption, many of which surfaced during the conference, including:
- Organization: CSPs need to modify internal organizations because NFV will affect all departments. If you don’t believe us, ask your counterparts in the enterprise: retaining individual silos of technology is not sustainable. As a result, CSPs will need to develop or acquire the right competencies to make this transition. The operators presenting agreed that, without the proper organization in place, deploying NFV could end up costing more. .
- Lack of standards: Standardization of NFV architecture is still in process. However, operators clearly told attendees they could not wait for the normal time period required for standardization to occur. A specific area where standards are required is in the APIs that the various components should provide to achieve real interoperability. Operators do not want to face large expenses for system integration and related uncertainties. Another example is the format of the VNF images: Today, there are too many different formats provided by the VFN vendors.
- Service orchestration: CSPs see this as their biggest challenge due to undefined scopes and interfaces as of today, but they need it now. Consensus seems to be that virtualization is not so difficult, but orchestration is!
- Hybrid networks: Operators know they will not move 100% of their network to NFV all at once; they recognize that they will have to manage partially virtualized and partially physical networks. However, this creates another problem: ensuring that operators have OSS tools capable of providing an end-to-end view across disparate domains. Operators need to know if their current network management application can support these hybrid environments.
It was clear from my time at the TM Forum that NFV is a hot topic for most operators. Indeed, many of them are eager accelerate their deployments, but are looking for standardized solutions to help them get there. And it also appears that many have learned from the enterprise’s adoption of virtualization technology that organization, culture and management tools are very important to a successful transition. Oh, yeah—and orchestration, too!
Is your company thinking about deploying NFV? Let us know your plans and let’s see how we can help you with your deployment. Empirix offers solutions to accelerate this transition and ensure service levels are met or exceeded.
About the author
Angelo is currently focusing on Product Marketing at Empirix for Customer Experience Assurance applications. He has worked in the Telecom sector for the last 25 years, with previous experiences in Software Development, Product Management, Solution Architecture and Strategic Marketing mainly focused on probes-based passive monitoring systems for CSP.