Agile Controller-DCN

To help customers adapt to cloud service changes in the Internet era, Huawei provides the CloudFabric solution for new-generation cloud computing data center networks. The solution aims to build flexible, virtual, and open cloud data center networks for customers and to support the efficient development of enterprise cloud services.

The Agile Controller-DCN is the core component of the Huawei CloudFabric solution. The Controller implements unified control and dynamic scheduling of network resources to rapidly deploy cloud services.

Specifications

Feature Description
Network Service Provisioning
  • Supports interconnection with the mainstream cloud platform OpenStack or third-party applications from Layer 2 to Layer 7. The cloud platform or third-party applications invoke the standard interfaces to provision network services.
  • Supports independent network service provisioning (including association with computing platforms) to implement automatic network deployment.
Fabric Management Uses the standard VXLAN protocol to implement automatic network deployment, including VXLAN protocol encapsulation. The Agile Controller-DCN also supports VXLAN Layer 2 and Layer 3 interconnection and interconnection between VXLAN and traditional networks.

  • Supports various VXLAN networking scenarios and management and control of software and hardware network devices.
  • It also allows hybrid access of multiple types of terminals such as physical servers, VMs, and bare metal servers in different scenarios.
Service Function Chain
  • Supports the IETF-based SFC model and adopts PBR or NSH as traffic diversion technologies to guide the service traffic to different nodes for service processing. In this way, the topology-independent SFC function with graphical orchestration and automatic configuration is implemented.
  • Provides VAS services, including security policy, NAT, and IPSec VPN.
O&M and Fault Location
  • Supports monitoring of physical, logical, and tenant resources.
  • Supports visibility of the application, logical, and physical network topologies. Mappings from the application to logical topology, and from the logical topology to physical topology can also be displayed.
  • Displays forwarding paths of VTEPs and VMs in VXLAN scenarios, implementing precise location from the logical network to the physical network.
  • Supports intelligent loop detection and provides one-click repair.
  • Supports detection of Layer 2 or Layer 3 network connectivity between VMs, as well as between VMs and external networks, through IP Ping and MAC Ping, helping administrators quickly rectify faults.
  • Supports traffic mirroring (traffic on VMs or bare metals can be mirrored to remote addresses through GRE tunnels).
Openness
  • Based on ONOS and compatible with ODL architecture.
  • Supports interconnection with the mainstream OpenStack platform (standard OpenStack, Red Hat, Mirantis, and UnitedStack) using northbound interfaces such as RESTful, RESTCONF, WebService, and Syslog from Layer 2 to Layer 7.
  • Supports interconnection with physical and virtual network devices using southbound protocols, such as SNMP, NETCONF, OpenFlow, OVSDB, JSON-RPC, and sFlow.
  • Supports interconnection with a computing resource management system, such as VMware vCenter and Microsoft System Center, for collaboration with network and computing resources.
Reliability
  • Adopts distributed cluster deployment. A single cluster supports a maximum of 128 member nodes. The service control node supports dynamic expansion without service interruption.
  • Supports deployment of cluster members in the same Layer 2 network or across a Layer 3 network as long as routes between cluster members are reachable.
  • Load balances the northbound cloud platform API requests or web access to different controller nodes.
  • Supports southbound load balancing capability. Devices on the entire network are evenly distributed for management by different controller nodes. If a fault occurs on one of the controller nodes, the network devices managed by it can be smoothly switched to other normal nodes to avoid service interruption.
Management Capacity and Performance Single-node cluster configuration

  • Number of managed physical network devices: 600
  • Number of managed physical servers: 3,000
  • Number of managed VMs: 60,000

Typical configuration: Three nodes

  • Number of managed physical network devices: 1,800
  • Number of managed physical servers: 9,000
  • Number of managed VMs: 180,000
  • VM online rate: 200 per second

Typical configuration: Five nodes

  • Number of managed physical network devices: 3,000
  • Number of managed physical servers: 15,000
  • Number of managed VMs: 300,000
  • VM online rate: 350 per second

Algunas de nuestras REPRESENTACIONES