NFV constructs convergent new network

New industry trends are driving operators to adopt IT technologies to restructure CT networks for a competitive edge

  As the mobile Internet becomes more widespread in the digital economic era, traditional telecom operators' voice and short message services (SMS) are continuously affected and squeezed by OTT APPs. The competitive environments and competitors that operators face are changing due to the thousands of OTT players that the telecom industry needs to compete with. user experience and behaviors are also changing as evidenced by Real Time, On Demand, All Online, DIY, and Social (ROADS). New competition and new user experiences and behaviors require that operators make changes to their business, marketing, R&D, operation, and service models. These changes require the support of an agile telecom network.

Traditional telecom networks are silo networks and capital expenditure (CAPEX) and operation expenditures (OPEX) are excessively high. This makes service innovations hard to deliver. These problems have become a major bottleneck for operators' development in the mobile Internet era.

Telecom network restructuring is the basis for delivering the user experience change and solving these problems. The major task for telecom network restructuring is to adopt IT technologies in order to restructure CT networks for a competitive edge. New technologies such as cloud computing bring new opportunities for operators' telecom networks and provide the technical basis for network restructuring. For traditional telecom network elements (NEs), software and hardware are decoupled. However, cloud computing technologies are used to share resources, flexibly allocate these resources, and deliver Network Functions Virtualization (NFV). This is known as cloudification. According to research done by the prominent consulting firm Infonetics, core network is the first step for a cloud-based telecom network.

Efficient software architecture and design ensure cloud-based core network capabilities

As the switching and control center for a telecom network, a cloud-based core network needs to continue to satisfy carrier-grade service level requirements, such as high reliability, high performance, and fault self-healing. The cloud-based core network also needs to be open. This requires that core network possess efficient software architecture and design.

Cloud-aware architecture

Cloud-aware software architecture is essential to core network capabilities. Cloud software architecture must ensure that telecom software is deployed by layer (such as service distribution, service processing, and data storage) in order to separate service logic and session data. A stateless design is adopted for service processing units. The following figure illustrates the typical architecture:

                                                                                                                                                                                                                            

                                                                                              Figure1 Cloud-aware architecture

According to the architectural design, session data in communications is separated from service logic. Dedicated distributed database modules are used to store such data. Programs (i.e. service processing units) only process service logic and do not need to store the session data. At the front end, service distribution modules perform load distribution. Since service processing units do not store session data, service distribution modules do not need to distribute messages to specific processing units according to user/session, therefore, flexible distribution strategies are supported. For example, messages can be distributed according to the load and health conditions of processing units.

A direct benefit of cloud-aware architecture is flexible scalability of service processing units. Virtual machine resources can be flexibly added or released according to the traffic fluctuation. For scalability, service processing units do not need to consider sessions being processed. Scalability can be completed very quickly. The traffic will be automatically rebalanced in minutes.

Carrier-grade service level assurance

As the center of telecom network, core network must satisfy carrier-grade service level requirements. The following common technical methods can be used to ensure service level agreements:

  • Multi-layer reliability strategies
  • KPI-based health check/fault self-healing
  • Service queue traffic control
  • High performance design based on virtual computing/storage/network resource optimization
  • Heterogeneous resource pools

For the cloud-based core network, software and hardware are decoupled and software is separated at different layers, therefore, reliability strategies need to be deployed from the bottom layer (servers, storage, and networks) to top-layer APPs. Methods such as multiple network adapter redundancy, network adapter bundling, multi-disk array networking, and APP-layer HA are utilized to ensure 99.999% reliability of the entire system.

KPI-based health check/fault self-healing is also a major method for ensuring carrier-grade service level agreements. For traditional fault processing, the system can be switched only after the entire system stops working and only overall system switchover is supported. This entails huge risks. The KPI-based health check monitors service KPIs in real time and compares them with historical key KPI statistics. When key KPIs deteriorate to a preset threshold, alarms are generated and an automatic recovery is completed rapidly based on fault causes

Service queue traffic control does not depend on CPU usage. The resource load, including the CPU, dynamic memory, and message packages, is considered as the delay for invoking messages in the message queue. When resource usage is high, the number of service messages invoked from the traffic control queue is reduced to increase the service message delay and quantity in the queue. When the service message quantity or delay exceeds the threshold, messages overflow from the queue to trigger traffic control. This helps reduce the impact that CPU fluctuation and inaccurate monitoring have on traffic control and improves CPU usage.

For the virtualization technology used by NFV, performance deteriorates because scheduling management and interrupt for virtual machines consume a large number of system resources. Performance is even reduced by 20%-30% in some extreme scenarios. To ensure the high performance of core network software, the software design must be optimized. The optimization can be achieved from aspects of virtual computing/storage/network resources. Virtual computing performance can be improved through resource consumption reduction of virtual machine scheduling and interrupt and improvement of memory access efficiency. Virtual storage performance can be improved through raw device mapping (RDM). Virtual network performance can be improved through the reduction of IO interrupts, and reduction of data copying between the OS kernel mode and user mode (DPDK), etc.

The cloud-based core network also needs to consider coexisting with a core network consisting of traditional hardware architecture. These two types of core networks cannot be separated; therefore, heterogeneous networking is adopted for these two types of core networks to form a unified resource pool. This allows O&M and services to be processed consistently. This is also a very important capability.

Openness

Dedicated software and hardware and closed service systems for traditional telecom networks complicate service on-line and impede innovation, therefore, operators cannot effectively compete with OTT players. The cloud-based core network primarily aims to build an open network.

An open cloud-based core network needs to meet the following conditions:

  • Support COTS hardware and multiple Cloud OSs.
  • Be able to pre-integrate with multi-vendor software/hardware.
  • Deliver an open, collaborated, and innovative industrial eco-system to aggregate industry values.
  • Support open source code.

In the future, the system also needs to support a mixture of COTS hardware and Cloud OS. That is, one cloud-based core network can be deployed on a heterogeneous data center formed by multiple COTS hardware and Cloud OS.

Huawei CloudCore solution leads core network cloudification

Huawei CloudCore solution fully complies with the software architecture and design requirements of a cloud-based core network and it covers all core network NEs. Furthermore, Huawei CloudCore solution also contributes a lot to NFV standards, has gained industry recognition, and has made multiple contributions to both industrial eco-system establishment and business cases.

                                                                                          

                                                                  

                                                                                                            

                                                                                                     Figure2 Huawei CloudCore Architecture

For NFV standards, Huawei has joined all NFV industrial organizations and open source communities. Huawei joined the ETSI NFV working group and is the co-chairman of the NFV architecture group. Huawei is involved in all nine research topics of the working group and has made important contributions to a number of topics. 1153 proposals have been accepted by the NFV organization, which makes Huawei as No. 1 contributor.

Huawei's dedicated efforts have also been acknowledged by the industry. In the 2014 IMS World Forum, Huawei CloudCore solution won the Most Innovative Virtualized IMS Solution award due to its leading technology, successful pilots, and commercial experience. According to the latest report released by the prominent consulting firm Current Analysis, at the end of 2014, Huawei was selected by global operators as the best SDN/NFV solution provider for two consecutive years.

On January 20, 2015, as a NFV leader, Huawei officially launched the industry-leading NFV Open Lab in Xi'an China. The NFV Open Lab has three primary responsibilities:

1. Develop multi-scenario and multi-vendor integration and verification capabilities. In other words, flexibly build multi-vendor integration and verification platforms for typical service scenarios. This includes gradually building a big data analysis platform through continuous integration and project practices in order to provide reliable data support for operators' NFV network planning and decision-making.

2. Make joint development and quick innovations and perform mutual qualification and authorization with industrial organizations, operators, and partners.

3. Build an open and cooperative NFV industrial eco-system. More than 150 partners have pledged their support to this project till now.

Huawei CloudCore solution has delivered multiple successful commercial cases. On July,2015, Huawei worked with Vodafone-Italy and successfully delivered the first cloud IMS commercial project in the industry. In this project, Huawei provided cloud IMS based on third party servers and Cloud OS. As the primary integrator, Huawei was responsible for the vertical integration work for the entire cloud-based system from the bottom layer (COTS hardware and Cloud OS) to the top-layer IMS software. By utilizing their professional delivery team, rich integration experience, and efficient integration tools, Huawei completed the delivery for commercial use within a short period.

With leading architecture and design, great industrial contributions, industry acknowledgement, and mature commercial capabilities, Huawei CloudCore solution will provide strong support to operators in order to build new, agile telecom networks and lead the cloudification journey for the core network.