Hello everyone,
Today I will share with you network reliability design.
In a large scale network , link/node failures can happen and affects the operation of the network. Huawei recommends the deployment of high availability feature to minimize the impact of failure on the services. The high availability features can be divided the four levels: physical device level, link level, protocol level and service level. The following sections discussed the respective high availability feature.
1.1 Device Level HA Design
Huawei adopts the carrier-class hardware and software design. The main principle of the design is as follows:
l Key hardware component adopts redundant backup design. The key component such as controller board (e.g. MPU), forwarding board (e.g. SFU), power modules and fans. Key hardware component are connected over passive backplane and are hot swappable.
l Control plane and forwarding plane are separate. As such, the forwarding plane can continue to forward traffic even control plane is experiencing active/standby switchover.
1.2 Link Level HA Design
Physical link level reliability involves designing redundant link interconnection between two devices. It also can protects against board failure causing both link to go down at the same time.
Usage of Eth-trunk means binding several physical links as a logical link and share the same IP Address. The traffic over the trunk interface will load-balance among several physical member links. When one member link is down, other links within same Eth-trunk interface will take over the traffic.
Using eth-trunk between the NE-X16A and NE-X8A links. It is easy to extend in future.
Configuration template for Eth-Trunk
NE | [router]interface Eth-Trunk X [router-Eth-TrunkX] mode lacp-static [router] interface GigabitEthernet X/X/X [router-GigabitEthernetX/X/X] eth-trunk X |
1.3 Protocol Level HA Design
NSR is a type of reliability technology that keeps the neighbor relationships of a device during the active/standby switchover of main control boards on the device.
The advantages of NSR are as follows:
NSR on the local device does not depend on or affect the remote device. Therefore, the local and remote devices can communicate properly.
The route convergence speed of NSR is higher than that of NSF.
NE | [Router] switchover mode nonstop-routing |
After enabling NSR, the protocol can be GR helper with GR configuration.
1.4 BFD
To minimize the impact of a fault on services and improve network availability, a network device must rapidly detect communications faults between adjacent devices so that the upper layer protocol can resolve the issue and recover services.
BFD provides the following features:
Low-cost fast fault detection for channels between adjacent forwarding engines. Faults can be detected on interfaces, data links, and forwarding engines.
A single mechanism capable of real-time detection over any media, at any protocol layer.
the BFD can deploy on these scenario:
BFD for ISIS
BFD for VRRP
BFD for TE LSP
BFD for TE Tunnel
That is all I want to share with you! Thank you!
