Cloud Fabric Data Center Network Solution
Enable Data Center Networks to Be More Agile for Cloud Services
Background and Challenges
A data center is an integrated Information and Communications Technology (ICT) application environment resulting from data concentration. By combining data computing, network transmission, and storage, a data center functions as the most critical infrastructure for enterprise business operations. In an era of cloud computing, enterprises and carriers have begun focusing on how to build data center infrastructure that supports sustainable cloud service development.
Cloud computing is capable of assisting enterprises in lowering Operating Expense (OPEX) and improving Operation and Maintenance (O&M) efficiency. By leveraging cloud computing technology, enterprises can maximize returns while minimizing resources. According to Forrester Research, a world-renowned research and advisory firm, global cloud computing applications are being carried out on a grand scale. Market share for cloud computing applications will reach US$241 billion by the end of 2020, up from US$4.07 million in 2011. The global market of Infrastructure as a Service (IaaS) is expected to garner US$5.9 billion in total revenue by the end of 2014. In addition, the global market share of Software as a Service (SaaS) will skyrocket from US$21.2 billion in 2011 to US$92.8 billion by the end of 2016. Meanwhile, the ecological web of cloud computing services matures day by day. For example, 400,000 or more commercial accounts, including The New York Times and the National Association of Securities Dealers Automated Quotation (NASDAQ), have already deployed Amazon Web Services (AWS) in over 190 countries and regions across the globe.
Cloud computing technology uses an innovative computing model that allows users to obtain nearly infinite computing capability and a wide variety of information services through networks at any time. This innovative business model provides on-demand computing and services, with charges based on quantity. While cloud computing applications are scaling up, cloud computing services continue to advance data centers towards Big Data, virtualization, and the Internet. As a result, cloud computing services pose new challenges for cloud data centers and their network architecture.
Trend 1: Big Data Leads to Exponential Network Traffic Growth.
Big Data is a massive, diversified information asset with a high rate of growth. Stronger decision-making and insight as well as higher process optimization capabilities are achieved by leveraging new processing models. Big Data is characterized by huge volume, broad variety, high velocity, and significant value. There are four typical Big Data service architectures: distributed architecture, server cluster, parallel computing, and social media application. Recent years have seen unprecedented development for Big Data. According to the International Data Corporation (IDC), a famous market research, analysis, and advisory firm, total Internet traffic will increase fourfold by 2015. Gartner, the world's leading information technology research and advisory firm, predicts that the average data traffic volume of enterprises around the globe in the next five years will grow to 800% of current levels. Gartner also forecasts that the Average Annual Growth Rate (AAGR) of global data center IP traffic will be 33%, and the world's total data will reach 4.8 ZB by 2016.
As Big Data results in exponential traffic growth, 10GE servers in data centers have become mainstream servers, and interfaces connecting data center networks have evolved to 40GE/100GE. A data center's internal traffic is absolutely the main data traffic. To adapt to this trend, data centers have an urgent need for flattened, scalable network architecture that provides non-blocking, high-speed packet forwarding capability.
Trend 2: Virtualization Demands Strong Resource Integration Capability.
Virtualization logically abstracts and uniformly represents ICT resources, and plays a significant role in the construction of large data centers. In 2013, millions of cloud computing users carried out various services using a virtualized cloud computing platform. Server virtualization proportion has increased to more than 70%, and network function virtualization has become an industry trend.
In this era of cloud computing, enterprises require that data centers have a stronger virtualized resource integration capability to improve resource use efficiency and collaboration efficiency. Virtualized architecture of data centers encompasses server virtualization, storage virtualization, network virtualization, and network Value-Added Service (VAS) virtualization. Functioning as the bearer of all ICT resources, a data center network must integrate various computing, storage, and network resources in the data center to implement on-demand utilization and scheduling of ICT resources.
Trend 3: Accelerated Internet Innovation Demands Programmable Networks.
Information Technology (IT) service innovation has new requirements for Communications Technology (CT) networks. For example, Google conducts customized development of network forwarding paths, increasing its bandwidth use efficiency from 30% to nearly 100%. In another example, a large Internet company in China wants to build larger and more reliable networks by self-defining and optimizing network protocols inside a data center.
[Icon representing a green, sustainable approach] [Icon representing a house or building]
Traditional vs. SDN Networks
Functions of traditional data center networks are hardware-based and evolve slowly. To deploy new services and functions, existing network devices must be replaced. The existing network devices cannot quickly adapt to the new services. Consequently, service provisioning efficiency is low, and the commercial use speed of networks is far slower than that required by the new services. How do we make an enterprise ICT system more competitive? How can we quickly introduce new services and functions? How do we enable network administrators to focus more on user experience and service innovation without the constraint of complicated and tedious network device faults? The answer is Software Defined Networking (SDN), that is, programmable network.
The Big Picture: Huawei Cloud Fabric Data Center Network Solution
To help its customers quickly adapt to changes in cloud computing services, Huawei has put forth an innovative, next-generation cloud computing data center. The aim of Huawei Cloud Fabric is to build an elastic, virtualized, and open cloud data center network for customers and to provide support for the sustainable cloud service development of enterprises. Huawei CloudEngine (CE) series data center switches lead the industry in high-performance data center switches. Based on Huawei's next-generation Versatile Routing Platform version 8 (VRP8), Huawei CE provides ample features for data center services. Agile Controller is Huawei's cloud data center network controller, which can uniformly control and schedule ICT resources and quickly deploy cloud services. Huawei Cloud Fabric supports multiple mainstream cloud platforms in the industry and can bear a broad variety of cloud services and applications. The Huawei Cloud Fabric Data Center Network Solution applies to the Internet, financial, and energy industries, government sector, large enterprises, and carriers.
Diagram illustrating Huawei Cloud Fabric architecture: Cloud Apps (VMware, FusionSphere, vCenter, System Center, OpenStack) connect to Cloud Fabric via Agile Controller. The fabric includes Controller zone, Network zone, Service/Security zone, Server zone, Storage zone, Management zone, Active data center, Standby data center, and DCI interconnection. Solutions include Fabric Solution, SDN Solution, DCI Solution, Security Solution.
Huawei Cloud Fabric Data Center Network Solution Sub-Solutions:
- Intra-data center connection solution
- Inter-data center connection and disaster recovery solution
- Data center security and application optimization solution
- Data center network management solution
Diagram showing network management and security/application optimization solutions for Active Data Center and Disaster Recovery Center. Components include Egress router (NE40E), NMS (eSight), Core/Aggregation layer (CE12800/CE7800), Access layer (CE6800/CE5800), Servers & Storage (Blade Servers, Rack servers, TOR, IP SAN, FC SAN), IP/MPLS, OTN/SDH. The diagram also shows Cloud Fabric in both data centers.
Customer Benefits
Huawei Cloud Fabric leverages an industry-leading architecture that can help customers build an elastic, virtualized, and open data center. This architecture also provides an agile solution innovation capability, adapts to fast-changing cloud services, and supports continuous cloud service development.
Huawei Cloud Fabric: Industry-Leading Architecture, Continuously Innovative Solution
- Elastic: Features CE12800 with highest switching capacity (64 Tbit/s), upgradable to over 100 Tbit/s. Supports TRILL (Layer 2 network with 512 nodes) and CSS/SVF (virtualization, simplified management). EVN for Layer 2 network expansion across data centers.
- Virtualized: Offers 1:16 virtualization capability, core switch reuse. Supports Nv03 for 16M multi-tenant expansion and vFW for 4K virtual firewalls. nCenter platform for VM migration and policy management.
- Open: Includes ENP Line Card for programmable forwarding plane. Agile Controller for unified control and dynamic scheduling. Cloud Ecosystem for cloud computing data center integration.
Timeline graphic from 2011 to 2014, highlighting the evolution of SDN, DCN, DCI, and the emergence of Cloud Fabric.
Elastic Fabric: Supporting Flexible Expansion of Cloud Services More Agilely
Huawei Cloud Fabric provides three-layer elastic scalability: device-level, system-level, and data center-level, satisfying high-speed development requirements for enterprise cloud services.
Diagram illustrating three layers of elastic scalability: Device-Level (features like 64 Tbit/s switching capacity, 128x100GE ports), System-Level (TRILL with 512 nodes, SVF+CSS for single management node), and Data Center-Level (EVN for integration across 32 data centers, 32K tenant services, 256K VM migration). The overall scalability is stated as 30 times the industry average.
Device-Level Elastic Scalability:
Huawei Cloud Fabric provides data center network products with super large switching capacities and continuous, seamless capacity expansion. Huawei CE12800 series data center core switches offer the industry's highest switch capacity (64 Tbit/s), upgradable to over 100 Tbit/s. A single line card has T-bit forwarding capacity. The CE12800 supports 12x100GE line cards. Huawei USG9500 series data center security gateways feature an "NP + multi-core + distributed" architecture, supporting 160Gbit/s application-layer firewall throughput per card and 1 Tbit/s device throughput, with up to 960 million concurrent connections.
System-Level Elastic Scalability:
Huawei Cloud Fabric supports Transparent Interconnection of Lots of Links (TRILL), enabling large Layer 2 networks with over 512 nodes, supporting over 18,000 10GE servers. It combines Huawei's proprietary Cluster Switch System (CSS) and Super Virtual Fabric (SVF). CSS virtualizes multiple core switches into one, while SVF expands heterogeneous core switches vertically, virtualizing Leaf switches into remote line cards of a Spine switch. This simplifies management, improves reliability, and maximizes forwarding efficiency for east-west traffic.
Data Center-Level Elastic Scalability:
Huawei Cloud Fabric provides efficient, three-layer connections among data centers for Layer 2, Layer 3, and Storage Area Network (SAN) services. It supports Ethernet Virtual Network (EVN) for cross-region service expansion across 32 data centers over IP WAN, integrating resources into a large IT pool. This enables server clusters and smooth migration of 256K Virtual Machines (VMs). Leveraging Huawei 400G core routers and 100G WDM products, it offers a reliable "IP+OTN" disaster recovery solution.
Virtual Fabric: Implementing Dynamic Deployment of Virtual Resources More Agilely
Huawei Cloud Fabric fully supports data center virtualization, including IT resource, network resource (device, network, VAS), and VAS resource virtualization, holistically integrating ICT resources to improve O&M efficiency and reduce TCO.
Diagram illustrating Network Resource Virtualization and IT Resource Virtualization. Network Resource Virtualization shows Device Resource (Campus, Office, Production, DMZ, CE12800 with 1:16 network core), Network Resource (Tenant 1, 2, n, Network resource pool, 16M multi-tenant transmission), and VAS Resource (vFW1, vFW2, vFW3 with 4K virtual firewalls). IT Resource Virtualization shows DC1 and DC2 with Network policy migration, VM, VM migration at any location, and VM migration on any platform.
IT Resource Virtualization:
Huawei Cloud Fabric provides diversified Ethernet Fabric technologies for large Layer 2 network transmission across data centers, enabling VM migration and easy IT resource scheduling. It supports dynamic collaboration with VM platforms, allowing network policies to migrate 10 to 20 times faster than the industry average.
Network Resource Virtualization:
Huawei Cloud Fabric enables in-depth virtualization of network resources, allowing secure sharing among multiple services, departments, and tenants. Using Virtual System (VS) technology, Huawei CE12800 switches offer 1:16 device virtualization capability. It supports multi-tenant network solutions like NVO3/VXLAN/NVGRE and TRILL. The CE12800 acts as a virtualized gateway supporting up to 16M tenants. Huawei Cloud Fabric also provides a VAS resource pool, with Huawei USG9500 supporting 4K virtual firewalls and Intrusion Prevention System (IPS).
Open Fabric: Accelerating Cloud Service Innovation More Agilely
Huawei Cloud Fabric implements full-scale network openness with an open fabric, an open controller, and an open ecosystem, enabling agile adaptation to cloud service innovation and quick provisioning of new services.
Diagram showing Open Fabric architecture: Cloud Apps (VMware, Microsoft System Center, FusionSphere) connect via Open API to Agile Controller. Agile Controller connects to CE12800 and Cloud Fabric. Open Controller details Northbound interface (connects to platforms via APIs) and Southbound interface (manages devices via protocols like OpenFlow). Open Fabric details Control plane (connects to platforms/controllers) and Forwarding plane (programmable ENP).
Open Fabric:
Huawei Cloud Fabric offers full-scale forwarding plane and control plane programmability. Huawei CloudEngine switches support programmable line cards using proprietary Ethernet Network Processor (ENP) chips for software-defined network functions and rapid expansion. It provides open APIs (Python/XML/Restful/OMI) and protocols (OpenFlow) for connecting to third-party controllers. This enables customers to innovate and deploy new services four times faster than the industry average.
Open Controller:
Huawei Agile Controller is a cloud data center network controller integrating multi-tenant network management, service chain orchestration, and VM access control. It provides unified control, dynamic scheduling of ICT resources, and quick cloud service deployment. Its northbound interfaces connect to mainstream cloud platforms, and southbound interfaces use Open APIs and OpenFlow to control Huawei and third-party devices, enabling network customization and seamless convergence of networks and services.
Open Ecosystem:
Huawei actively builds a cloud computing data center ecosystem through its Cloud Ecosystem plan. It partners with IT device vendors for cloud platforms, virtualization, servers, and storage. Huawei collaborated with Microsoft on a Hybrid network virtualization solution using CE12800 switches. Huawei is a Gold member of the OpenStack Foundation, promoting SDN industry chain development.
Product Portfolios
The Huawei Cloud Fabric Data Center Network Solution provides multi-layered network products.
Visual representation of product categories: Routers (NE40E Series), Switches (CloudEngine 7800/6800/5800, CloudEngine 12800), Security Devices (USG6600, USG9500/AntiDDoS, SVN), OTN (OSN 1800, OSN 8800).
Cloud Computing Data Center Products Key Features:
- 12x100GE/24x40GE high-speed line cards
- 64 Tbit/s switching capacity
- 4 Tbit/s bandwidth per slot
- T-bit throughput security protection capability
- Defends against 100+ security attack types
- Hierarchical and integrated security protection
| Network Location | Device Model | Description |
|---|---|---|
| Data center switches | CE12800 series | Industry-leading data center switches |
| CE7800 series | High-density 40GE core/aggregation switches | |
| CE6800 series | High-density 10GE access box switches, connecting to 40GE uplink devices | |
| CE5800 series | High-density GE access box switches, connecting to 10GE or 40GE uplink devices | |
| Data center security products | USG9500 | High-performance firewalls |
| USG6000 | Next-Generation Firewalls (NGFWs) | |
| AntiDDoS series | DDoS attack defense | |
| Routers | NE40E series | Egress routers for data centers |
| Optical Transport Network (OTN) products | OSN series | OTN devices for data centers |
| Data center intelligent NMS | eSight | Visible intelligent network management platform |
| Data center network controller | Agile Controller | Data center network controller |
Why Huawei?
Huawei is proudly backed by 20 years of experience in the IP field and an outstanding series of network products and solutions. Recognized as one of the world's leading network solution providers, Huawei has an excellent long-term plan for network development and a firm determination to invest in the network field. Most importantly, Huawei's world-class research capabilities and experts offer unparalleled experience with network standards and chip development.
As a member of ONF, IETF, and IEEE, Huawei participated in SDN standards research, contributing greatly in areas of network migration to SDN, including product development and improvement in customization capability. Huawei remains committed to providing intelligent, programmable, and open networks to customers through its accumulated carrier-grade network experience and innovative products.
For more information, please visit: http://enterprise.huawei.com/.
