In Ultra-broadband era, with the remarkable increase of new services such as VoIP, HDTV and mobile, internet traffic grows much faster than that of single-chassis router processing capability. Single-chassis routers have physical restrictions in capacity, reliability and expansibility and have evidently been a focus of network expansion.

The router cluster technology has been adopted to address the above issues. The cluster system consists of a group of routers that are interconnected with each other through the cluster central chassis (CCC) and acts as a single router logically. Cluster system effectively deals with network traffic explosion, fosters the flat network architecture, and protects operators' investment. Although Cluster routers have above advantages, they request higher reliability and higher stability.

The key challenge of cluster is how to realize the smooth expansion. Traditionally, operators generally replace component of smaller capacity with one of higher capacity to increase network traffic. That will interrupt the service and increase the ultra investment of operators. Meanwhile, cluster routers must switch control plane quickly to ensure higher reliability of system when network has fault.

In addition, venders now generally connect the CCC and cluster line-card chassis (CLC) in full-mesh topology. It is easy for cluster to be deployed and be managed, and meets the operator' requirement at specific phases which is the compact-capacity clusters (1+N or 2+N) scenario. While in case of rapid bandwidth increase and cluster expansion, the N2 problems caused by full-mesh connection will increase OPEX dramatically.

To address the above challenges, Huawei provides innovative in-service hardware expansion (ISHE) solution based on the leading hardware and software design. The ISHE solution ensures seamless expansion of clusters and takes full advantages of original system hardware, which secures investment and reduces capital expenditure (CAPEX) and OPEX to the most extent for operators.

1. Flexible Hardware Architecture Design

CCC adopts the mode of backplanes in the vertical direction. Electric cross units (ECUs) are cross connected to SFUs in the form of an orthogonal matrix. As Figure1 shown:

 

The orthogonal matrix architecture realizes the Full-mesh connection between the data plane and switch plane. Data transmitted along each fiber can be hashed to each interface and then be balanced over all exchanging planes.

 

Figure2 shows an example that traffic was loaded on every switch planes. All Optical Flexible Cards in Electrical Cross Unit will receive traffic (optical signal) sent from CLCs. ECU change optical signal to electrical signal, then load balance these electrical signal to every switch plane of all SFUs. In this way, cluster improves the switch capacity farthest. In addition, innovative hardware architecture perceives cluster scale intelligently, configures switch mode and switch granularity dynamically, deploys OFC on demand. This architecture of flexible matrix ensures smooth expansion, hardware can be reused, which cuts down CAPEX and OPEX and protects investment as much as possible.

 

2. Innovative Software Platform Design

Huawei innovative software platform provides optimal solutions to various cluster deployments.

In compact-capacity clusters (1+N or 2+N) scenario, which includes a small number of CCCs and CLCs, the control planes of CCCs and the control planes of CLCs are connected in dual star topology to achieve Full-mesh. It is easy for cluster to be deployed and maintained through dual star topology. It achieves high performance-price ratio for compact-capacity clusters.

In large-capacity clusters (4+N or 8+N) scenario, which includes a great number of CCCs and CLCs, the dual star topology, however, makes the clusters more complexity and harder maintenance. Backed up by extensive R&D capabilities, Huawei provides the innovative control plane pool (CP-POOL) solution. Though this solution, all control planes of the cluster system are connected by a 10G fiber to form a RRPP (Huawei patent) ring.

 

1. Control planes with high reliability ensured by patented RRPP technologies: All MPUs are connected to an RRPP ring (comprising a main ring and a sub-ring). When a link of the cluster fails, the system automatically detects the fault and the control signaling enables traffic to be switched to an inverse link in 50 ms, which ensures high reliable operation of control planes.

2. Smart election of the "brain" for a cluster：The key concept of the CP-POOL solution is "pool".

In CP-POOL, all Main Process Units (MPU) of CCCs are integrated into a pool, where each MPU can participate in election. The winner automatically becomes the cluster master MPU. The real-time control information of master MPU will be synchronized to the slave MPU. When the master MPU fails, the slave MPU automatically becomes a new master MPU. Meanwhile, a new slave MPU is elected from the pool.

Huawei's innovative ISHE solution, which combines the flexible hardware and intelligent software, realizes cluster’s smooth expansion without changing hardware or interrupting service. As the bridge connecting single-chassis era to cluster era, the ISHE solution enables network core nodes to stride forward to the cluster era in the industry.

With All-IP convergence strategy, Huawei has been well-positioned to fully support All-IP network transformation for operators. As the core platform of Huawei’s IPTime solution for backbone networks, over 1,000 sets of NE5000E, including 40 clusters have been shipped to over 80 global operators, including China Telecom, China Unicom, Deutsche Telekom, France Telecom, Portugal Sonaecom and Singapore SingTel, as of June 2009.