The Fusion of IT and Telecom
The rapid growth in Internet usage from mobile devices is having a huge impact on the telecom industry. To serve this growth, telecom operators are working to improve efficiency through dramatic changes such as centralized procurement, network sharing and service outsourcing. Traditional network construction, however creates bottlenecks when trying to reduce expenditure and at the same time increase revenues. While Internet service providers can utilize low-cost commodity hardware and open source software to build their software-based infrastructure, telecom operators build their networks using dedicated hardware, which is expensive and greatly reduces service flexibility and deployment speed. It is imperative that telecom operators think about the transformation of their networks. The foundation for these transformations is an efficient and open platform, which allows both IT systems and telecom services to be run in a cloud.
A telecom cloud platform must be open
A large-scale survey of telecom operators’ plans for Network Functions Virtualization (NFV) revealed that the major factors driving cloud platform deployment plans are the desire to reduce network total cost of ownership (TCO) and increase network flexibility through open interfaces (APIs). Having an open cloud platform is the key to realizing these benefits.
NFV envisions a cloud platform which distributes the functionality of traditional telecom "boxes" and uses virtualization software to allow telecom service software to move from specialized hardware to commodity hardware, thereby reducing costs and increasing service deployment efficiency. This move to virtualization is hampered by the current market situation, where a small number of vendors dominate the market. Open virtualization software and an open cloud computing platform are required by operators to avoid vendor lock-in while minimizing the investment in virtualization and cloud platforms. An open cloud platform will also provide standard and open management APIs and service APIs, facilitating telecom operators' service deployment and simplifying resource and service management.
The first step in moving from traditional IT hosting, a business already provided by many operators, is to offer Infrastructure as a Service (IaaS) and Software as a Service (SaaS) to enterprises. Telecom operators have advantages in providing these two services, such as reliable network and data center infrastructures and strong after-sale support capabilities. In addition, an increasing number of governments require data to be retained locally, giving telecom operators an advantage in developing regional cloud services.
An open cloud platform not only helps telecom operators overcome the challenges posed by software and platform costs, it also can provide standard service APIs which will allow fully open hybrid clouds, enhancing the value of cloud services for enterprises.
FusionSphere open cloud platform
OpenStack is an open source project supported by major IT and telecom companies, along with developers from all over the world. The goal is to build a versatile cloud platform that is easy to deploy and scale. The platform is oriented to public and private clouds and provides an open and standard cloud operating system for data centers. About 850 organizations from 132 countries are involved and Huawei, a gold member of the OpenStack Foundation, has been deeply involved since it was founded, providing vision and contributing key technology to the efforts.
Utilizing the open and standard interfaces provided by OpenStack, an open cloud computing platform with excellent flexibility can be built, reducing the costs of cloud data center deployment and management. To meet the unique needs of ICT convergence and telecom operators, Huawei created FusionSphere, an OpenStack-based cloud operating system.
Latency is a key performance indicator for carrier-class applications. Telecom applications are typically composed of multiple virtual machines that communicate frequently with one another. Compared with dedicated physical hardware, typical enterprise-class virtualization adversely affects the communication performance of applications. Based on innovative technologies, FusionSphere hypervisor improves the communication performance of VMs by 1.5 to 3 times, greatly reducing the latency, and enabling traditional carrier-class applications to be smoothly migrated to the cloud platform.
To meet the requirements of IT and communication applications, FusionSphere provides reliability mechanisms independent of applications, such as high availability (HA) and lightweight fault tolerance (FT). High performance, high reliability, and low latency enable a better user experience in the environment of network function virtualization (NFV).
An indication of FusionSphere’s server virtualization hypervisor optimization is that it out-performs enterprise-class virtualization in the SPECvirt_sc2010 tests.
Currently, many different storage systems exist in data centers are closely linked with the applications using the storage. Much of this storage is direct attached storage (DAS) located in servers, which is not virtualized and is not managed and scheduled in a unified manner. FusionSphere adopts SDS(software defined storage), an innovative distributed storage system which can be deployed on general servers, and it helps the local storage form a resource pool very close to the CPUs. Compared with traditional disk arrays, SDS increases the IOPS of applications by 2 to 3 times, improves the throughput by 6 times, reduces the latency between server and storage to one fifth to a half of the original one. Thus FusionSphere SDS provides better performance for mission-critical applications and helps telecom operators maximize return on investment.
Distributed cloud data centers
FusionSphere allows organizing virtualized infrastructure resources into unified logical resource pools and provides centralized resource management and scheduling across physical data centers, creating a super-large cloud data center housing millions of virtual hosts. By supporting OpenStack APIs, FusionSphere supports an open architecture which is compatible with heterogeneous physical devices and virtualization platforms. Key technologies, such as fine-grained resource scheduling that crosses data centers and SDN-based traffic engineering, deliver higher resource efficiency and better energy efficiency, along with ensuring the QoS of services and reducing the total cost of ownership (TCO).
FusionSphere uses an overlayed virtualization network component that is based on SDN (Software Defined Network) and VXLAN to construct a full mesh Layer 2 network in and across data centers, simplifying application deployment and making resource scheduling flexible. Being aware of topologies and QoS, the SDN controller dynamically selects the correct routing path and adjusts the WAN routing policy and bandwidth among multiple data centers, reducing cross-data-center link bandwidth cost by 60% to 70% on average. The complete SDN consists of virtual switches, service gateways that provide virtual firewall and VPN functions, and centralized SDN controller clusters. SDN supports Layer 2 to Layer 7 network features, high availability of VMs and live migration across data centers.
By providing automated network configuration to support application/VM deployment, FusionSphere significantly simplifies application deployment and allows flexible resource scheduling.
By merging the cloud computing concepts of Internet Players with the telecom network, not only can traditional telecom services be delivered more efficiently, but the network can also allow IT resources to be delivered as services, opening up new revenue opportunities for operators. Huawei’s FusionSphere is designed for ICT convergence and will help telecom operators smoothly migrate IT and telecom services to cloud platforms, delivering benefits both from providing cloud services to enterprises, as well as from supporting the transformation of networks and services. FusionSphere is a proven platform which has been deployed by telecom and enterprise customers in over 40 countries.
— Gu Jiongjiong, Chief Architect of Cloud Computing, Huawei & Zhang Xiaosong, Product Manager of Cloud Computing, Huawei