ZTE ZXR10 T8000 Cluster Router

Current Perspective

ZTE's ZXR10 T8000 multichassis core router is competitive in the IP core router market because it addresses the capacity and scale requirements of the next wave of IP core (bearer) networks. ZTE launched its new T8000 at P&T/Wireless & Networks Comm China 2009 and since then has announced deployments for China Telecom, China Mobile and China Unicom as well as Smart Telecom, Atheeb, ECT, and VNPT. The T8000 is based on the vendor’s in-house-designed networking silicon, which allows service providers to build very large router configurations that can collapse multiple network layers. As service providers look to scale their networks, cluster-router architecture has become one of many tools that can be used to scale capacity and port density and simplify network topology.

    The T8000 router is based on ZTE’s ZXRIC silicon chipset, which delivers up to 3.84Tbps switching capacity per chassis and supports 40 GigE interfaces. The three-chipset includes a scalable forwarding engine (ZXRIC PFE) that supports 40G (100G future) wire-speed forwarding, a support chip (ZXRIC SF600 and SA) that supports switching access to a three-stage 600Gbps switch fabric, and a multipolicy traffic management chip (ZXRIC TME) that provides fine-grained control over network traffic with five levels of QoS. T8000’s massive capacity comes from its ability to cluster multiple chassis into a single logical router. Interconnecting multiple racks allows for linear equipment capacity extension. The system currently supports multiple clustering methods, for example, 1+4 mode and 2+8 mode. Ultimately, the ZXR10 T8000 will support a 16+64 mode configuration, capable of delivering up to 200Tbps of switching capability and has up to 2048 40G interfaces or 1024 100G interfaces. Switch fabric options such as 1+1, 2+1 and 3+1 are available to meet specific resiliency requirements. The T8000 is also designed to comply with emerging national and international initiatives in green telecom. To reduce overall power consumption, it uses an intelligent and dynamic power management scheme for power supplies, line cards, and other system components.

   ZTE entered the IP super core router market somewhat later than its competitors but has gained significant traction with major service providers. The launch of T8000 has been well received, especially by ZTE’s service provider customers China Unicom and ETC, and enables ZTE to compete better with vendors such as Cisco and Juniper. They have all announced and been delivering multichassis solutions for some time.

Buying/Selecting Criteria Manageability

   T8000 and other ZTE service provider routers are managed through the Netnumen U31 management system, which handles the needs of the element, network and service layers. The system can manage multiple network layers, namely, the access layer, bearer layer, control layer and application layer. This means the solution can be deployed in different O&M scenarios. Netnumen provides full FCAPS functionality.

  The Netnumen U31 service management function includes a TE manager, QoS manager, multicast manager, and VPN manager. The service management functions help simplify service delivery by coordinating data forwarding resources, handling faults, and providing a visual perspective of QoS.

   Netnumen U31 has the ability to operate, maintain, and manage multiple NEs in a centralized way. It supports different types of OS and DB and includes topology management to provide a functional diagram of the whole network, including all nodes and links. It also provides fault analysis—such as alarm delay, alarm count, alarm restraint, alarm merger—and provides privileged access control and log records.

   External interfaces include northbound interfaces such as TL1, Syslog, CORBA, SNMP, XML and MML. Southbound interfaces include SNMP (V1, V2C, V3)—standards used in various products and private MIB—TELNET, MMl, and TR069.

Performance and Architecture

   T8000 currently provides 3.84Tbps full duplex switching capacity per chassis/rack, giving it strong performance metrics on a chassis and rack basis. After the ZXRIC 100G chipset is released, the switching capacity per rack will double to 7.68Tbps. The system’s 36-RU chassis footprint permits only one chassis per rack. Its current packet-forwarding performance is 1600 million packets per second for the 40G system.

   T8000 uses a distributed forwarding and a non-blocking cross bar switching fabric with three-plane separation (forwarding, control, and management) architecture. Separation of the forwarding plane from the management and control planes is a key architectural attribute necessary for next-generation service delivery.

   T8000 supports between 4 and 10 million IP routes (RIB) based on configuration. These levels are more than adequate for projected Internet growth in the foreseeable future. Scalability for all routers in this class ranges from 1 to 6 million routes.

   T8000 supports multichassis configurations in order to achieve massive scale, and multiple combinations of central fabric chassis (CFC) and line card chassis (LCC) can be configured. For example, a back to back mode is offered, 1+2, 1+4, 2+2, 2+4, 2+8, 4+16 and 16+64 mode. In terms of the existing design, the T8000 supports at most a 16+64 mode cluster configuration with a total switching capacity of 200Tbps.

Physical Specifications and Certifications

   T8000 is rack-mountable, using standard 19-inch telco cabinets. Operators can deploy a single chassis version by using the LCC, which includes a main processing unit, switch fabric unit, packet forwarding and physical interface units. Service providers can deploy one 36-RU LCC per cabinet.

   T8000 can be expanded to include multiple chassis for greater scale. Cluster configurations consist of a CFC and multiple LCCs, and this provides a maximum capacity of 200Tbps. Cluster mode configurations include non-redundant CFC configurations (1+2, 1+4) and redundant CFC configurations (2+2, 2+4, 2+8, 4+16, 16+64) of CFC and LCC respectively.

   T8000 employs an intelligent power supply management system that automatically controls and optimizes power consumption of each module according to traffic conditions. The fan speed is variable, and the fan system intelligently initiates the line card and goes into hibernation to further reduce power consumption.

  The T8000 draws approximately 5100 watts of power and weighs 210kg when fully configured. Its power consumption is one of the lowest in this product class. System power can be provided by DC or AC power modules, rated at 8000W and 2000W respectively for either chassis type (LCC, CFC). The -48V DC supplies are configured in 1+1 redundancy mode, and the 220V AC is configured in 4+4 redundancy mode.

   T8000 is designed to comply with key safety and environmental standards such as CE, FCC, UL, RoHS and Anti-9 magnitude earthquake. The system has not been NEBS certified, which is critical for deployment in facilities of most North American service providers.

Port Density and Scalability

   T8000 supports up to 768 GigE ports per chassis. GigE ports are standard for such platforms, and ZTE’s current GigE port densities range from ten to 48 ports per physical interface module. This supports combinations of 10/100/1000Mbps.

   T8000 currently supports up to 64 10 GigE ports per chassis using the 40G packet forwarding unit (PFU). Up to 128 10 GigE ports will be supported with the future PFU-100. The system currently supports multiple versions of the physical interface comprising a one, two or four-port 10GigE WAN Ethernet optical; a one, two or four-port 10GigE LAN Ethernet optical; and a one or two-port 10GigE LAN/WAN interface.

   T8000 currently supports up to 16 40 GigE ports via a single port line card using the PFU-40 packet forwarding unit. Up to 32 40 GigE ports will be supported with the future PFU-100.

  T8000 will support up to 16 100 GigE ports via a single port line card using the future PFU-100 packet forwarding unit. The 100G ZXRIC chipset needed for the PFU-100 is expected to be generally available in 2012 following trial production in 2011.

  T8000 supports POS interfaces so that carriers can deploy up to 16 OC-768c/STM-256 or 64 OC-192c/STM-64 ports per chassis. POS support has also includes OC-48c/STM-16, OC-12/STM-4 and OC-3/STM-1. SONET interface requirements are still strong in the IP core and will be for the next two to three years as ethernet interfaces continue to gain dominance in the market.

   Unlike the Juniper T-series and Cisco XR12000 series routers, T8000 does not support ATM interfaces. ATM support is diminishing as a requirement in next-generation networks.

Routing and Service Features

   T8000 provides Layer 2 support such as MAC address management, VLAN, Q-in-Q, Super VLAN, Smartgroup and interface binding. Effective Layer 2 services enable the service provider to transport non-routed traffic through the core network in a cost-effective way.

   T8000 provides Layer 3 support including IPv4 unicast, IPv4 multicast, IPv6 unicast, IPv6 multicast, MPLS and TE. It also supports MPLS L2/L3 VPN, 6vPE, MPLS-TE, MPLS static tunnel, GRE, IPSec, and DS-TE. Layer 3 service support enables the service provider to support routed traffic, such as video, which requires multicast/unicast and caching for quality of experience and transport efficiency.

   T8000 supports a full range of IPv4 and IPv6 services, including interworking. As the industry transitions from IPv4 because of the shortage of new addresses, IP core routers will need to provide effective migration support for both forms of IP traffic.

    T8000 provides a full complement of quality of service (QoS) functions, such as classification, label, traffic policing, congestion control, queue scheduling, shaping, QPPB and H-QoS.

   T8000 supports ongoing O&M functions such as CLI, GUI (Netnumen U31), MPLS VPN NM, QoS NM and TE NM, Ethernet OAM, MPLS OAM and an SLA tool.

Service Assurance

   T8000 has separate data, control/management, and monitoring planes. The data plane supports wire-speed packet processing and non-blocking forwarding. The control and management planes, which can be redundantly configured, manage system configuration and processes for protocol and signaling. They also generate and maintain routing tables and facilitate information interaction between all components. A separate monitoring plane detects abnormalities with the power, fan and temperature and raises alarms.

     T8000 uses a distributed and modular operating system, ZXROS that separates software processes so that interactions that could negatively impact the normal running of the system are avoided. The T8000 software consists of multiple logical units, such as MPLS, where each logical unit is supported by 1+1 backup and can be run on different MPUs. The distributed protocol implementation framework enables the system to be scaled up to massive capacity while retaining resiliency and redundancy and maximizing available system resources.

   T8000 uses a system controller (SC) to manage and control the system and provide multichassis resiliency. The SC is a logical component that can run on any main processing unit (MPU) of cluster system to achieve 1+1 redundancy between different chassis.

   T8000 supports in-service component upgrades (ISCU), which enables online upgrades without affecting service delivery. Each component has a backup process that can be upgraded independent of the other. The system also supports a hot patch process to immediately solve problems as they occur.

  Resiliency features include non-stop routing (NSR), non-stop forwarding (NSF), traffic engineered fast reroute (TE-FRR) and label distribution path FRR (LDP-FRR). The system also supports graceful restart (GR) and pseudowire redundancy. Given the requirements for “always on” services, system resiliency is a key attribute of any core router.

  T8000 also provides clock synchronization services in order to support jitter and delay-sensitive traffic that transits through the IP core network. The system supports IEEE 1588 with synchronous ethernet interfaces, out-of-band 1PPS+TOD interface (where 1588v2 is not required), in-band ethernet interfaces (where service and time information are transmitted via one interface), and adoptive SSM or BMC protocols (to realize automatic protection switching of time link to guarantee reliable transmission).