H3C S12500R Data Center Flagship Converged Switch Router

H3C S12500R is a switching router product launched by New H3C Technology Co., Ltd. (hereinafter referred to as H3C) for WAN, 5G bearer network and DCI interconnection scenarios in data centers. Forwarding performance and very rich wide-area traffic scheduling features.

The S12500R is a switch router product with excellent performance in the industry. A single machine can provide up to 3072 line-speed 10G/25G ports or 768 line-speed 40G/100G/400G ports, providing ultra-high-density 10G, 25G and high-density 40G, 100G, 400G Capability; In the face of the burst characteristics of WAN traffic, the "distributed ingress cache" technology is innovatively adopted, which can realize 200ms data cache and meet the requirements of burst traffic in IPRAN, DCI and other network scenarios; at the same time, it supports independent control engine, detection Engine and maintenance engine provide powerful control capability and 50ms high reliability guarantee for the system.

S12500R products include S12500R-2L, S12500R-2XL, S12504R, S12508R, S12516R, S12508CR, S12516CR and other models, which can adapt to the port density and performance requirements of different network scales, and provide a strong equipment guarantee for wide-area interconnection construction. At the same time, combined with H3C series routers, switches, security, iMC and SDN solutions, it provides a full range of solutions for wide-area convergence and interconnection scenarios.

Advanced CLOS+ multi-level and multi-plane switching architecture

* Adopt advanced CLOS/CLOS+ multi-level and multi-plane orthogonal switching architecture, with no backplane/no midplane design, providing continuous bandwidth upgrade capability.

* Support high-density 10GE/25GE/40GE/100GE/200GE/400GE Ethernet ports to fully meet data center applications and future development needs.

* Independent switching network board, control engine and switching network board hardware are independent of each other, which maximizes equipment reliability and provides guarantee for continuous upgrade of subsequent product bandwidth.

* Dynamically variable cell fragmentation is strictly non-blocking switching, which improves the overall forwarding performance.

Comware V9 Containerized OS

The S12500R adopts the new-generation operating system Comware V9 independently developed by H3C. Compared with the previous-generation operating system, on the basis of integrating rich network features, the S12500R has a further open architecture and modular software architecture, supports containerized deployment, and can carry third-party software applications. .

* Rich network features: Comware has rich basic device functions, network functions and management functions, while Comware V9 provides comprehensive customization and tailoring capabilities: Linux infrastructure (Linux function modules, Docker capabilities), network functions, management functions ( SNMP, NetConf, CLI...) can be tailored).

* Openness and programmability: The native Linux kernel is used to facilitate kernel upgrades, and at the same time, it has better openness (it is more convenient to integrate open source Linux software into COMWARE and users to use third-party software), provide the ability to run third-party software seamlessly, and provide The programmable interface opens up programmability and supports user-defined network services.

* Containerization: It supports containerization, integrates Docker, and can deploy Comware in Docker containers and run containerized Comware or third-party programs.

Smart Connection Based on SRv6

* SRv6 is a future-oriented new-generation protocol type, which naturally supports IPv6 and satisfies access to massive address space; SRv6 can identify applications and tenants, realize intelligent routing such as delay and bandwidth, and ensure SLA; at the same time, SRv6 realizes a unified protocol, simplifying configuration.

* SRv6 uses segments with a length of 128 bits to define network functions, and then by arranging the segments, a series of forwarding and processing behaviors of network devices can be implemented to complete service orchestration. Compared with the MPLS SR protocol, it has stronger scalability and better compatibility with SDN controllers, which is more conducive to deploying applications in DCI, MAN and other scenarios.

* The notable feature of SRv6 is that the forwarding plane adopts IPV6, which makes it easier to achieve interconnection between different networks based on the reachability of IPV6. SRV6 forwarding is adopted within a domain, while between domains, only ordinary IPv6 forwarding is required, and MPLS is not required. Similarly, due to the need to convert MPLS to IP, do a lot of complex configuration.

Building Flexible Ethernet Based on FlexE Technology

* As a technical architecture based on Ethernet and industrial chain expansion, FlexE technology completely reuses the existing IEEE 802.3 Ethernet physical layer standard, and through lightweight enhancements at the MAC/PCS logical layer, it realizes flexible multi-rate interfaces and is compatible with IP The technology realizes seamless connection. Under the IP/Ethernet technology system, it better meets the needs of large bandwidth, flexible speed and channel isolation, which is in line with the development trend of technology and industry. The rise of services such as video and 5G, as well as the improvement and functional enhancement of FlexE technology, are accelerating the formation of the FlexE industry chain.

* Based on FlexE, 5G network segmentation can be realized, which can meet the bearing requirements of different services through the segmentation of network resources, and ensure the SLA of services (such as bandwidth, delay, etc.). Different services (such as eMBB enhanced bandwidth, autonomous driving, uRLLC and massive IOT interconnection, etc.) can be carried in the same IP network.

* Based on FlexE to implement hard-pipe private lines for major customers, through hard-pipe technology in the IP network, end-to-end rigid pipes can be constructed through the FlexE channelization function for the bearer of important private lines and low-latency-sensitive services. In the statistical multiplexing IP network, this end-to-end FlexE hard-pipe private line can provide service quality assurance for specific high-value customer services on the basis of making full use of the existing network infrastructure.

* Realize IP+Optical flexible networking through FlexE. As the UNI interface between routers and optical transmission network equipment, FlexE can realize the difference between the actual data stream bandwidth carried by the UNI interface and the bandwidth of the optical transmission network NNI interface WDM link through rate matching. One-to-one correspondence, which greatly simplifies the mapping of the FlexE interface of the router to the transmission equipment of the optical transmission network, and reduces the equipment complexity, investment cost (CAPEX) and maintenance cost (OPEX).

* FlexE Bonding is essentially a "L1 LAG (Link Aggregation)" technology. Based on the FlexE bonding technology, a link with a larger bandwidth can be constructed through interface rate combination. The problem of unbalanced traffic distribution across multiple links in traditional LAGs can achieve 100% balanced bandwidth allocation.

High Precision Time Solution 1588v2

* IEEE 1588v2 is a master-slave synchronization system. During the synchronization process of the system, the master clock periodically publishes the PTP time synchronization protocol and time information, and receives the timestamp information sent by the master clock port from the clock port, and the system calculates accordingly. Master-slave line time delay and master-slave time difference, and use this time difference to adjust the local time, so that the slave device time keeps the same frequency and phase as the master device time. IEEE1588v2 can realize frequency synchronization and time synchronization at the same time, and the accuracy of time transmission mainly depends on the accuracy of the frequency of the two condition counters and the symmetry of the link. Compared with traditional timing technology, IEEE1588v2 has obvious advantages. It adopts two-way channel, the precision is ns level, the cost is low, and it can adapt to different access environments and so on. Under the background of the industry that requires increasing precision, IEEE1588v2 has become an inevitable trend of development.

Innovative multi-engine control design

* Adopted innovative hardware design, through independent control engine, detection engine, maintenance engine for the system to provide powerful control ability and high reliability guarantee.

* Independent control engine, providing powerful main control CPU system, easy to process all kinds of protocol packets and control packets, and support protocol packet fine control, providing the system with perfect ability to resist protocol packet attacks.

* The independent Detection engine provides Fast Fault Detection and Restoration (FFDR) system with high reliability and performance. It is used for Fast Fault Detection such as BFD and interworks with protocols on the control plane to support Fast protection switching and Fast convergence. Fast fault detection is implemented to ensure service continuity.

* Independent Maintenance engine, intelligent EMS (Embedded Maintenance subsystem-embedded Maintenance Subsystem) CPU system, the CPU system supports intelligent power management, can support board sequential power on and off (reduce the power impact brought by board power on at the same time, improve the life of the device, Reduce electromagnetic radiation, reduce system power consumption), check the online status of equipment.

* An independent monitoring engine, which is completely separated from the service control plane, monitors hardware working status in real time, including power load and power adjustment, fan speed automatic adjustment, and dynamic power consumption allocation.

Highly available M-LAG architecture

* S12500R series switches support Multi-chassis Link Aggregation Group (M-LAG) cross-device Link Aggregation technology (original DRNI technology). By virtualizing two physical devices into one device at the forwarding level, cross-device Link Aggregation is realized, and the control level is independent from each other. This improves board - level reliability to device - level reliability.

* M-LAG Decouples the control plane from the data plane. Devices on each node can be independently upgraded to ensure zero service interruption.

* Active-active device access provides device-level redundancy protection and traffic load balancing while improving system reliability.

Ensure data center level reliability

* The S12500R series products provide a dedicated FFDR system for fast fault detection, such as BFD, and implements interworking with protocols on the control plane, supporting fast protection switching and fast convergence.

* Supports BFD for VRRP, BGP, IS-IS, RIP, and OSPF static routes.

* Supports NSR, GR for OSFP, BGP, and IS-IS.

* The control engine and SFU hardware are independent of each other to realize the physical separation of the control plane and forwarding plane, and the control engine is 1+1 redundant; N+M SFU redundancy; Redundant design of fan system; Power modules are N+M redundant. To maximize the fault isolation capability and reliability of the system.

Multi-level security protection

* Multilevel protection and security of the control plane. The control plane policy feature of the S12500R manages the packet flow from the data plane (DP) to the control plane (CP) by configuring QoS filtering and rate limiting. In this way, the S12500R switch can identify and protect important packets and discard invalid packets under DDoS attacks. This ensures that the control plane maintains normal forwarding and protocol status under attack or heavy traffic.

* Supports massive ACL rules and full-speed forwarding; It implements detailed secure access control over various L2/IPv4/IPv6/MPLS packets and their field combinations.

* Supports hardware-level encryption Macsec technology (802.1AE). Unlike traditional soft encryption technology with end-to-end basic application layer protection, Macsec protects and manages control protocols for bridging networks and other data through cryptographic techniques that identify data sources, protecting information integrity and providing reprotection and confidentiality services. By confirming the frame sent by this station, the Layer 2 protocol can be protected from attack.

Distributed cache mechanism and refined QoS

* Innovative adoption of "distributed entry caching" technology for next generation data center traffic surges. Each port can precisely allocate bandwidth and shape traffic for all traffic flowing to the port. The precise scheduling of the forwarding plane ensures distributed caching in the Ingress direction, effectively sharing and utilizing the cache space distributed on all line cards, and providing better caching effect.

* The network application model has been transformed from C/S to B/S model. The change of application mode leads to the increase of network burst traffic, and the large cache mechanism has become an urgent need for network equipment. The S12500R supports burst traffic of 1600ms per 10-gigabit port. Combined with the distributed inlet cache mechanism, the S12500R meets the requirements of large data centers for high burst traffic.

* A single chip supports 8GB cache, and a line card supports a maximum of 32GB (4 x 8GB, each chip is independent and cannot be shared).

* The system supports a maximum of 64K hardware queues, fine QoS and traffic management, and assigns different priorities and queues to different users and traffic flows as required to ensure different bandwidth, service delay, and jitter performance.

Comprehensive maintenance and detection mechanism

* Online status detection mechanism, through the dedicated maintenance engine, can realize the switch network board, backplane communication channel, service communication channel, key chip, memory, etc. If a module is faulty, it is reported to the system through the EMS.

* Board isolation: The specified board is isolated from the forwarding plane and does not participate in the forwarding plane. However, the isolated board is still in the control plane and can be managed. The system can perform real-time diagnosis and CPLD upgrade on the board without affecting the system services.

* Support Ethernet OAM, providing a variety of device-level and network-level fault detection means.

Open Application Architecture

* S12500R series products are designed based on the concept of OAA (Open Application Architecture), and an innovative business platform Open to the outside world is launched.

Green design

* The S12500R series supports intelligent power management through the intelligent EMS engine system. It can power on the boards in sequence (reducing the power impact caused by the simultaneous power-on of the boards, improving the service life of the devices, and reducing electromagnetic radiation), control the power off of the boards, isolate faulty or idle boards, and reduce the system power consumption.

* S12500R series fans adopt high efficiency PWM speed control fan, support stepless speed control. The system automatically collects the temperature of the board, calculates the fan speed curve based on the actual situation, and sends the command to the fan enclosure. The system supports fan status monitoring (such as speed monitoring and fault alarm). The system automatically adjusts speed in zones based on the ambient temperature and board configuration, reducing device power consumption and running noise, effectively reducing ambient noise, and prolonging the life of fans.

* S12500R series products support automatic detection of internal ports. When no interface board is configured for a slot or no cable is connected to the port, the system automatically closes the corresponding internal ports, saving power consumption.

* The minimum power consumption of a 10G port is less than 3.4W, a 40G port is less than 10.4W, a 100G port is less than 13.8W, and a 400G port is less than 20.3W

* S12500R series products adopt front and rear straight ventilation, strict front and rear air duct design, high ventilation and heat dissipation efficiency, energy saving and environmental protection, can meet the requirements of efficient heat dissipation and energy consumption in the data center room.

* the S12500R series uses a single spu 50mm high design, that is, service slot 4 is 200mm (< 5U), service slot 8 is 400mm (< 10U), and service slot 16 is 800mm (< 20U).