Cloud computing and network functions virtualization (NFV) can help manage rapid demand growth while reducing capital and operational expenditures (CAPEX and OPEX). No wonder service providers are paying attention. These savings significantly lower service providers’ total cost of ownership (TCO) and increase agility — critical to thriving in today’s challenging telecom environment. Virtualizing applications also simplifies complex processes, such as healing, scaling, and software upgrades, providing further agility and flexibility. Such benefits can increase customer satisfaction and reduce churn, which may prove to be the tipping point when service providers evaluate whether to move certain applications to a cloud platform.
NFV and innovation in operations Much has been said about how virtualization and the cloud may be used in the telecom industry to improve the infrastructure and operations TCO. Initial efforts centered on virtualization’s ability to optimize hardware. Lately, the focus has shifted to operations. While virtualizing some network functions will undoubtedly bring CAPEX savings, NFV’s greatest contribution will be that it enables a new way of approaching telecommunications. This means much more than just optimizing inefficiencies inherent in current processes. Service providers can — and should — take advantage of NFV technology to redefine their current operations. This will require 3 major steps:
- Map out every current process in detail
- Analyze what can be automated (that is, handled by an NFV platform) to reduce complexity
- Redesign operations to be much simpler and more agile
Cost drivers for NFV
Many parameters may be considered when developing a business case to analyze the impact of migrating an application to NFV (Figure 1). There are 3 categories of cost drivers:
- CAPEX: one-time investments in fixed assets with a useful life extending beyond the taxable year
- Infrastructure OPEX: ongoing costs directly related to the infrastructure (e.g., maintenance)
- Process OPEX: ongoing staffing costs directly related to the daily management of activities or processes required to provide services or applications
6 areas of cost saving enabled by NFV
1. Capacity growth Traditional approaches to adding capacity follow a 4-step process (Figure 2) to deploy a new server infrastructure.
The NFV deployment process (Figure 3) differs from the traditional process in a number of ways.
Costs for deploying NFV are slightly higher initially due to the professional services needed during deployment. However, this is a one-time cost. As the service provider becomes more familiar with the infrastructure, it will likely use its own operators and perform these tasks in house. With NFV, applications can share infrastructure, so the service provider’s operations team will only need to be familiar with a very limited number of infrastructure elements. In succeeding years, total server replacement and growth process costs are greatly reduced. NFV’s virtual scaling and automated application deployment capabilities reduce capacity growth process costs significantly. 2. Software upgrades Today, upgrading with both new programmed software releases and ad-hoc patches follows 4 phases:
- Obtain the new software
- Test the new software
- Install and configure
The last phase generally consumes the most time and resources. Introducing an NFV platform doesn’t typically change the way the provider plans and obtains software. However, NFV offers a reduced timeline and lower costs to stage tests and create environments. Service providers can use “sandbox” testing environments without dedicated equipment. This lets them create simplified test cases which can be executed in parallel, and reduces testing time by about one-third. NFV simplifies installation and configuration. Traditionally, service providers open maintenance windows at night to install and configure a predefined number of servers individually. With NFV, the service provider can upgrade 4 servers per night in a 5-hour maintenance window. The lead time maintenance window grows over time as the service provider increases the number of physical servers to keep up with growing traffic needs (Figure 4).
NFV changes the whole process. The total number of servers is no longer relevant for installation and configuration. Application recipes are used to push upgrades automatically, in a matter of minutes, to all servers in parallel. This automation provides dramatic gains in agility. 3. Healing process Device failures can result in loss of service for many users and increase churn. To reduce this risk, service providers traditionally deploy fully redundant architectures. This costly security buffer requires double the amount of physical infrastructure, with much of it standing idle. A device failure is not the only issue that can require a healing process. Service providers also need to be able to address OS failures, application failures and distributed denial of service (DDOS) attacks. Traditional healing process Today’s healing process consists of 3 stages:
- Issue identification
- Trigger and execute solution process
- Perform ‘Post-mortem’ root cause analysis (RCA)
Lead times to identify and solve a problem vary depending on the issue at hand. It tends to be simpler and faster at the hardware and operating system layers, whereas actually solving an issue tends to be faster at the application layer. DDOS attacks are the fastest both to identify and solve, but tend to consume more of the operations team’s time because they are so common. RCA is performed by operators once service continuity has been assured. Identifying the root cause of a problem allows service providers to make the changes necessary to avoid reoccurrences. NFV healing process With NFV, devices run as virtualized functions and are protected by the self-healing properties of the hypervisor and orchestration layer. The healing process is fully redefined as the business continuity process is decoupled from the problem itself. To provide end-to-end application resiliency and reliability, NFV platforms incorporate mechanisms for automated healing, based on the monitored infrastructure and application-level KPIs. When failures occur, the system automatically creates a new instance with the same specifications to ensure application availability at all times. By simplifying the healing process and developing a simple solution using automated virtual scaling capabilities, NFV can significantly reduce healing costs. 4. Floor space, power and cooling Real estate, power and cooling are OPEX infrastructure costs. They are directly related to the number and characteristics of physical infrastructure items managed for a specific deployment. Provided all constants remain equal, reducing physical hardware will lower the total costs of real estate, power, and cooling by the same proportion. The main drivers for these costs are:
- Real estate: number and size of infrastructure items and square foot cost
- Power: rate of energy consumption and cost per kilowatt hour
- Cooling: a factor of 1:1 of power consumption
With NFV, real estate costs are reduced because the technology requires fewer physical infrastructure items. With traditional approaches, load balancers and other networking equipment such as switches are placed separately from servers. Power costs are reduced because NFV makes it possible for service providers to replace older servers sooner. Older servers consume about twice as much energy as new ones. Lastly, cooling costs are generally calculated as a 1:1 ratio to power costs, hence, cooling costs decrease in the same proportion as encountered with power. 5. Maintenance and software licenses Maintenance is also an OPEX infrastructure cost. It’s directly related to the number and characteristics of physical infrastructure items managed by the operations teams. Many traditional infrastructure elements require a yearly maintenance fee, including servers and the network equipment, such as load balancers, switches, and routing ports. While any chosen NFV system will have associated licenses and maintenance fees, there will be considerably fewer licenses than when using a traditional approach. This is because far fewer infrastructure elements are required, and NFV platforms can be shared between applications or services as capacity needs change. 6. Hardware infrastructure Virtualizing physical assets improves resource utilization by creating virtual machines, each with its own operating system on a single physical hardware asset. An NFV platform goes a step further. It enables dynamic placement of the virtual machines, which further improves hardware optimization. Traditional deployments operate in a “siloed” architecture. Servers are dedicated to one application, resulting in an inefficiently high number of servers. NFV enables a new model, where all underlying hardware forms a pool of resources shared by all the applications running on the same platform. Furthermore, the ability to share the infrastructure permits a new cost model. The cost of idle capacity should not be allocated to a specific application, but is available for other applications on demand. Service providers can expect significant reductions in server costs with NFV, since it uses far fewer servers than traditional approaches. More importantly, physical appliances, such as load balancers, can be eliminated. This article is excerpted from the Alcatel-Lucent strategic white paper entitled "Business Case for Moving DNS to the Cloud". To contact the author or request additional information, please send an email to firstname.lastname@example.org.
It’s all about speed and capacity. Mobile subscribers now have higher expectation levels than ever before and as mobile operators begin to deploy LTE-Advanced in earnest, subscribers will continue to demand more speed and more capacity. So the challenge for operators is clear – give the people what they want. This is where LTE Carrier Aggregation (CA) comes into its own. Useable spectrum is a finite resource and also a mobile operator’s lifeblood. As a result, we have now seen operators begin pushing for CA to help them make the absolute most of their available spectrum. CA is one of the most sought after features of LTE-A enabling operators to multiply bandwidth delivered to users by using radio resources over several carriers. This aggregation of larger swaths of spectrum increases bandwidth and in effect lets operators create fatter pipes that support services with a better Quality of Experience (QoE).
Aggregation across and within bands
One important application of LTE Carrier Aggregation is in helping operators to aggregate fragmented spectrum resources. Over time, operators acquire spectrum via auctions as well as through mergers or acquisitions of other operators. In these circumstances there often isn’t a perfect meshing and spectrum holdings may vary by market. CA lets operators aggregate these disparate chunks of spectrum spanning across different bands by supporting inter-band CA. This is the most common use for CA as most spectral assets that operators own have been acquired piecemeal over time. In some markets, many operators are also turning to LTE-TDD as a way of further augmenting capacity of existing LTE-FDD networks. These operators have large swaths of TDD spectrum (over 100MHz in some cases) and can use intra-band CA to combine several carriers to achieve higher speeds as a way of differentiating their services.
Other Uses of Carrier Aggregation
In addition to maximizing the use of available spectrum and exponentially increasing bandwidth, CA also enables operators to make better use of network resources through load balancing. Using real-time information including the network load across carriers as well as per user RF conditions, the network can intelligently and dynamically load balance traffic across multiple carriers, distributing resources more evenly and improving network utilization. The end result is an improved user experience with performance gains of between 30% and 70% depending on network load. CA is also useful for interference management. It can help reduce interference and improve network performance through the intelligent allocation of resources. In heterogeneous networks, it can provide better Inter-Cell Interference Cancellation (ICIC) by designating a specific carrier for the Physical Downlink Control Channel (PDCCH).
Today’s traffic flows are inherently asymmetric in nature with the majority of traffic being on the downlink – consider how many more people download videos than upload them. By bonding non-contiguous spectrum into a single, wider channel, operators can address the asymmetry of data flows between downlink (DL) and uplink (UL) channels. CA also allows operators to bond frequency bands together to provide an enhanced user experience at the cell edge. For example, operators can take advantage of the better propagation characteristics of low frequency bands to deliver better coverage and combine them together to also deliver increased capacity at the cell edge. A further benefit of CA in LTE-A is that operators can gain access to previously ‘stranded’ asymmetric spectrum not accessible by Release 8/9 terminals. Say for example an operator had acquired downlink only spectrum not useable by LTE, since it requires both an uplink and downlink channel. However, now with CA the operator can utilize the DL only spectrum as part of the multiple downlink carriers aggregated thereby recouping it for use within the LTE-A network. Again maximizing available resources and creating more capacity.
The way forward for LTE-A
For the reasons noted, Carrier Aggregation will become increasingly commercialized in the near future. It has the potential to make global roaming easier, since mobile devices will be able to support many different CA bands. Operators will also more than likely make use of CA to give themselves a competitive advantage over rivals, thanks to being able to advertise the highest speeds and throughput to customers. In a nutshell, Carrier Aggregation represents one of the most cost-effective and efficient way of addressing the capacity challenge and could be the biggest success of all LTE-A features. All indications are that we will see CA go mainstream in the not too distant future giving operators a truly valuable tool. To contact the author or request additional information, please send an email to email@example.com.