By
Raghu Kondapalli, Director of Strategic Planning at LSI
The ongoing convergence of video and cloud-based
applications, along with the exploding adoption of mobile devices and services,
are having a profound impact on carrier networks. Carriers are under tremendous
pressure to deploy new, value-added services to grow subscriber numbers and
increase revenue per user, while simultaneously lowering capital and
operational expenditures.
To help meet these challenges, some carriers are creating
some of these new services by more tightly integrating the traditionally
separate data center and carrier networks. By extending the virtualization
technologies that are already well-established in data centers into the telecom
network domain, overall network utilization and operational efficiencies can be
improved end-to-end, resulting in a substantially more versatile and
cost-effective infrastructure.
This two-part article series explores the application of two
virtualization techniques—software-defined networking (SDN) and network
function virtualization (NFV)—to the emerging unified datacenter-carrier
network infrastructure.
Drivers for virtualization of carrier networks in a unified
datacenter-carrier network
In recent years, user expectations for “anywhere, anytime”
access to business and entertainment applications and services are changing the
service model needed by carrier network operators. For example, e-commerce
applications are now adopting cloud technologies, as service providers continue
incorporating new business applications into their service models. For
entertainment, video streaming content now includes not only traditional movies
and shows, but also user-created content and Internet video. The video delivery
mechanism is evolving, as well, to include streaming onto a variety of fixed
and mobile platforms. Feature-rich mobile devices now serve as e-commerce and
entertainment platforms in addition to their traditional role as communication
devices, fueling deployment of new applications, such as mobile TV, online
gaming, Web 2.0 and personalized video.
Figures 1 and 2 show some pertinent trends affecting carrier
networks. Worldwide services revenue is expected to reach $2.1 trillion in
2017, according to an Insight research report, while the global number of
mobile subscribers is expected to reach 2.6 billion by 2016, according to
Infonetics Research.
To remain profitable, carriers need to offer value-added
services that increase the average revenue per user (ARPU), and to create these
new services cost-effectively, they need to leverage the existing datacenter
and network infrastructures. This is why the datacenters running these new
services are becoming as critical as the networks delivering them when it comes
to providing profitable services to subscribers.
Datacenter and carrier networks are quite different in their
architectures and operational models, which can make unifying them potentially complex
and costly. According to The Yankee Group, about 30 percent of the total
operating expenditures (OpEx) of a service provider are due to network costs,
as shown in Figure 3. To reduce OpEx and, over time, capital expenditures
(CapEx), service providers are being pushed to find solutions that enable them
to leverage a more unified datacenter-carrier network model as a means to
optimize their network and improve overall resource utilization.
Virtualization of the network infrastructure is one strategy
for achieving this cost-effectively. Virtualization is a proven technique that
has been widely adopted in enterprise IT based on its ability to improve
utilization and operational efficiency of datacenter server, storage and
network resources. By extending the virtualization principles into the various
segments of a carrier network, a unified datacenter-carrier network can be
fully virtualized—end-to-end and top-to-bottom—making it far more scalable,
adaptable and affordable than ever before.
Benefits of integrating datacenters into a carrier network
Leveraging the virtualized datacenter model to virtualize
the carrier network has several benefits that can help address the challenges
associated with a growing subscriber base and more demanding performance
expectations, while simultaneously reducing CapEx and OpEx. The approach also
enables carriers to seamlessly integrate new services for businesses and
consumers, such as Software-as-a-Service (SaaS) or video acceleration. Google,
Facebook and Amazon, for example, now use integrated datacenter models to store
and analyze Big Data. Integration makes it possible to leverage datacenter
virtualization architectures, such as multi-tenant compute or content delivery
networks, to scale or deploy new services without requiring expensive hardware
upgrades. Incorporating the datacenter model can also enable a carrier to
centralize its billing support system (BSS) and operation support system (OSS)
stacks, thereby doing away with distributed, heterogeneous network elements and
consolidating them to centralized servers. And by using commodity servers
instead of proprietary network elements, carriers are able to further reduce
both CapEx and OpEx.
Integrated datacenter-carrier virtualization technology trends
The benefits of virtualization derive from its ability to
create a layer of abstraction with the physical resources. For example, the
hypervisor software creates and manages multiple virtual machines (VMs) on a
single physical server to improve overall utilization.
While the telecom industry has lagged behind the IT industry
in virtualizing resources, most service providers are now aggressively working
to adapt virtualization principles in their carrier networks. Network function
virtualization (NFV), for example, is being developed by a collaboration of
service providers as a standard means to decouple and virtualize carrier
network functions from traditional network elements, and then distribute these
functions across the network more cost-effectively. By enabling network
functions to be consolidated onto VMs running on a homogenous hardware platform,
NFV holds the potential to minimize both CapEx and OpEx in carrier networks.
Another trend in virtualized datacenters is the abstraction
being made possible with software-defined networking, which is enabling
datacenter networks to become more manageable and more open to innovation. SDN
shifts the network paradigm by decoupling or abstracting the physical topology
to present a logical or virtual view of the network. SDN technology is
particularly applicable to carrier networks, which usually consist of disparate
network segments based on heterogeneous hardware platforms.
Technical overview of network virtualization
Here is a brief overview of the two technologies currently
being used in unified datacenter-carrier network infrastructures: SDN and NFV.
Software-Defined
Networking
SDN is a network virtualization technique based on the
logical separation and abstraction of both the control and data plane
functions, as shown in Figure 4. Using SDN, the network elements, such as
switches, routers, etc., can be implemented in software, virtualized as shown,
and executed anywhere in a network, including in the cloud.
SDN decouples the network functions from the underlying
physical resources using OpenFlow®, the vendor-agnostic standard interface
being developed by the Open Networking Foundation (ONF). With SDN, a network
administrator can deploy a new network application by writing a program that
simply manipulates the logical map for a “slice” of the network.
Because most carrier networks are implemented today with a
mix of different platforms and protocols, SDN offers some substantial
advantages in a unified datacenter-carrier network. It opens up the network for
incorporating innovation. It makes it easier for network administrators to
manage and control the network infrastructure. It reduces CapEx by facilitating
the use of commodity servers and services, potentially by mixing and matching
platforms from different vendors. In the datacenter, for example, network
functions could be decoupled from the network elements, like line and control
cards, and moved onto commodity servers. Compared to expensive proprietary
networking solutions, commodity servers provide a far more affordable yet fully
mature platform based on proven virtualization technologies, and industry-standard
processors and software.
To ensure robust security—always important in a carrier
network—the OpenFlow architecture requires authentication when establishing
connections between end-stations, and operators can leverage this capability to
augment existing security functions or add new ones. This is especially
beneficial in carrier networks where there is a need to support a variety of
secure and non-secure applications, and third-party and user-defined APIs.
Network Function
Virtualization
NFV is an initiative being driven by network operators with
a goal to reduce end-to-end network expenditures by applying virtualization
techniques to telecom infrastructures. Like SDN, NFV decouples network
functions from traditional network elements, like switches, routers and
appliances, enabling these task-based functions to then be centralized or
distributed on other (less expensive) network elements. With NFV, the various
network functions are normally consolidated onto commodity servers, switches
and storage systems to lower costs. Figure 5 illustrates a virtualized carrier network
in which network functions, such as a mobility management entity (MME), are run
on VMs on a common hardware platform and an open source hypervisor, such as a
KVM.
NFV and SDN are complementary technologies that can be
applied independently of each other. Or NFV can provide a foundation for SDN.
By using an NFV foundation combined with SDN’s separation of the control and
data planes, carrier network performance can be enhanced, its management can be
simplified, and new services can be more easily deployed.
***********************************
Raghu Kondapalli is director of technology focused on Strategic Planning and Solution Architecture for the Networking Solutions Group of LSI Corporation.
Kondapalli brings a rich experience and deep knowledge of the cloud-based, service provider and enterprise networking business, specifically in packet processing, switching and SoC architectures.
Most recently he was a founder and CTO of cloud-based video services company Cloud Grapes Inc., where he was the chief architect for the cloud-based video-as-a-service solution. Prior to Cloud Grapes, Kondapalli led technology and architecture teams at AppliedMicro, Marvell, Nokia and Nortel. Kondapalli has about 25 patent applications in process and has been a thought leader behind many technologies at the companies where he has worked.
Kondapalli received a bachelor’s degree in Electronics and Telecommunications from Osmania University in India and a master’s degree in Electrical Engineering from San Jose State University.
Intel to acquire LSI's Axxia Networking Business and related assets for $650 million in cash from Avago Technologies.
The new processors combine up to 16 ARM Cortex-A15 cores with LSI specialized networking accelerators to optimize performance and power efficiency. On-chip network accelerators deliver up to 50 Gbps packet processing, 20 Gbps security processing and 160 Gbps of Ethernet switching via 16 10G Ethernet interfaces. LSI plans to offer several versions with different core counts and throughput capabilities. The chip design leverages ARM’s CoreLink CCN-504 low-latency interconnect in 28nm process technology.
