.shock - Fotolia
NFV is changing the way service providers build and operate networks. Constructing virtual networks requires high-performance, reliable and scalable platforms -- NFV infrastructure or NFVi -- to support a wide range of telecom applications (virtual network functions or VNFs).
Telecom markets have a range of unique workload requirements that challenge NFVi suppliers. Telecom equipment can vary from large, highly scalable platforms that support millions of users and cost millions of dollars, to small access devices costing $1,000 or less each. In addition, telecom applications can be highly compute-intensive or require high-end I/O for bandwidth transport at low latency (some devices require both high-end compute and I/O).
In addition, the telecom industry has specific, unique requirements, including:
- High reliability: 99.999% uptime is required. Upgrades need to take place while the equipment is running or during very brief maintenance windows.
- Long life span: Typically hardware platforms are deployed for 5 to 7 years or more.
- Low latency: Sub-millisecond response times to support real-time applications like voice and video.
In addition, NFVi needs to support a broad software ecosystem of VNFs to deliver the required telecom-specific applications.
Telecom providers have many NFVi options
Commercial off-the-shelf (COTS) is shorthand for industry-standard servers (e.g., Intel and ARM), merchant semiconductors (e.g., Broadcom and Cavium), and standard operating systems (e.g., Red Hat and VMware). Advances in IT technology, including more powerful processors (e.g., Intel x86 and Cavium), faster switching fabrics (e.g., 40 GB), and advances in network software (e.g., SDN and NFV) have brought a wealth of network functions in scope for COTS. COTS is already in widespread use in telecom networks -- specifically, application servers, wireline and wireless core, OSS/BSS and, of course, service provider data centers.
An abundance of platform options for network and telecom suppliers exist, including:
Intel is the world's largest semiconductor manufacturer due to the popularity of its x86 architecture, which is used in the vast majority of PC and server products. Intel also offers network interface cards, Ethernet switching and highly reliable middleware (via its Wind River acquisition). Intel has recently introduced software and reference designs to promote the use of x86 (and SDN/NFV) in the networking/telecom industry.
HP is the largest supplier of server platforms, as well as a key provider of telecom software, security, storage and networking equipment, and offers extensive professional and support services for service providers. HP bundles infrastructure software (including multiple hypervisors, OpenStack and Wind River middleware) to deliver its NFVi platform. Telefonica recently selected HP as its NFVi platform.
Dell has a wide range of assets to support its NFV efforts, including servers, storage, software, networking and services. Dell partners with Red Hat to deliver OpenStack with its NFVi platform.
Cisco leverages its UCS server platform, virtualization and security software, and a wide range of network/telecom applications for its Evolved Service Platform, which supports NFVi. Cisco is in a large number of NFV trials and provides global professional and support services.
VMware is the leading supplier of virtualization software. It recently introduced vCloud for NFV, which provides multitenant isolation, multiple hypervisor support and open APIs. VMware positions vCloud for NFV as its middleware platform running on a wide range of standard servers.
Other NFVi suppliers offer middleware software (e.g., Red Hat and Wind River) and platforms -- including Huawei, Ericsson, Oracle, IBM, Lenovo, Radisys and Artesyn.
Advances in COTS technologies and the increased acceptance of virtual network applications (e.g., VNFs) are driving requirements for NFVi platforms. Service providers will increasingly leverage COTS platforms to deliver the performance and scalability required by their customers. NFVi suppliers will need to deliver high performance/scale and low latency, meeting telecom service providers' specific reliability needs. Development of a broad software ecosystem to support virtualization of telecom applications is also a key requirement for NFVi platforms.
What telecom infrastructure trends are driving the networks of the future?
How service providers can implement virtual network functions
Old meets new: Making SDN and NFV work with legacy equipment