OSPF Hub-And-Spoke Technology White Paper-6W100

HomeSupportResource CenterOSPF Hub-And-Spoke Technology White Paper-6W100
Download Book
Table of Contents
Related Documents

 

 

OSPF Hub-And-Spoke

Technology White Paper

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Copyright © 2021 New H3C Technologies Co., Ltd. All rights reserved.

No part of this manual may be reproduced or transmitted in any form or by any means without prior written consent of New H3C Technologies Co., Ltd.

Except for the trademarks of New H3C Technologies Co., Ltd., any trademarks that may be mentioned in this document are the property of their respective owners.

This document provides generic technical information, some of which might not be applicable to your products.

The information in this document is subject to change without notice.



Overview

Background

About hub-and-spoke networking

Currently, flat networks become popular in some industries. The hub-and-spoke network is a typical flat network. Compared with the traditional networks containing the core layer, distribution layer, and access layer, a hub-and-spoke network connects the core layer device (hub device) to access layer devices (spoke devices) without using the distribution layer, as shown in Figure 1. This networking approach reduces the number of devices and facilitates network management and maintenance.

Besides, the hub-and-spoke network delivers high scalability. To implement the complete three-layer architecture in the future, you only need to configure the core layer as the distribution layer and add new core-layer devices. Smooth network expansion is implemented with original network configurations reserved.

Figure 1 Hub-and-spoke network diagram

 

Solutions to the problems of using OSPF in a hub-and-spoke network

Typically, the hub device and the spoke devices in a hub-and-spoke network are configured to reside in the same OSPF area and the hub device establishes OSPF neighbor relationships with the spoke devices. If the hub device has a large number of OSPF neighbors, the following problems might arise:

·     How to keep the neighbor relationships stable?

Stable OSPF neighbor relationships are the operating basis of OSPF. If the hub device fails to receive or process hello packets, corresponding neighbor relationships may time out on the spoke devices and need to be reestablished.

·     How to shorten the network convergence time?

The more neighbors the hub device has, the longer time a network convergence operation takes. Therefore, the hub device is required to establish neighbor relationships and notifies the neighbors of the routing information changes as fast as possible.

·     How to avoid affecting other services?

Besides maintaining OSPF neighbors, the hub device also needs to be capable of supporting other services, such as another routing protocol.

To solve the above problems, H3C optimizes OSPF, enabling it to:

·     Give priority to the receiving and processing of hello packets.

·     Limit the number of neighbor relationships that can be concurrently established.

·     Ensure smooth packet transmission

·     Control LSA generation

·     Control the SPF calculation interval

Benefits

OSPF hub-and-spoke provides the following benefits:

·     Increase the number of supported OSPF neighbors, and improves the efficiency of neighbor establishment and LSDB synchronization.

·     Enable making configurations only on the hub device, thus facilitating network deployment and maintenance.

Implementation

Concepts

·     Hub device: The hub device is the core device in a hub-and-spoke network and is attached with multiple spoke devices. Besides supporting basic OSPF functions to ensure reachability at the network layer, the hub device processes services specific to the hub-and-spoke network.

·     Spoke device: Spoke devices are edge devices in a hub-and-spoke network, which are connected with user networks. Spoke devices only need to support basic OSPF functions to ensure reachability at the network layer.

Mechanism

Giving priority to the receiving and processing of hello packets

With this feature configured, OSPF gives priority to hello packets over other OSPF packets and other protocols' packets (such as BGP packets and IS-IS packets), ensuring the maintenance and establishment of OSPF neighbor relationships.

 Limiting the number of neighbor relationships concurrently established

During neighborship establishment, neighbors in the Exchange or Loading state exchange a large number of DD, LSR, LSU and LSAck packets. If many neighbors are in that state, too many packets will be exchanged, thus exhausting system resources. Consequently, a large number of packets may be discarded or may not be processed in time.

H3C enables OSPF to limit the number of neighbors in the Exchange or Loading state. If the total number of such neighbors exceeds the specified threshold, new neighbors cannot enter the Exchange or Loading state; instead, they will stay in the Exstart state. They can enter the Exchange or Loading state when the total number falls below the specified threshold. This optimization reduces the number of concurrently exchanged packets, and thus greatly reduces discarded packets and retransmitted packets, ensuring both fast and stable neighborship establishment.

Ensuring smooth packet transmission

In the hub-and-spoke network, the spoke devices exchange link state information through the hub device. The hub device also needs to notify its own link state information to the spoke devices. If there are many spoke devices, the hub device has to concurrently handle a great amount of packets generated due to neighbor state changes, periodical LSA updates and route flapping, which may overwhelm the system and thus affects neighbor establishment and the operation of other services.

To solve the problem, H3C introduces the smooth packet transmission function, which can decrease the forwarding rate of LSU packets on each interface. Thus, the system resources are less consumed, unnecessary retransmissions are avoided, the efficiency of neighbor establishment is improved, the number of neighbors supported is increased, and the normal operation of other services is ensured.

Controlling LSA generation

A router LSA describes the interface and neighbor information of a device within an area. Its length is in direct proportion with the numbers of the device’s interfaces and neighbors within that area, and thus becomes very large when there are many neighbors in that area. Each time a neighbor’s state changes to Full, a router LSA advertisement is sent to all neighbors. The LSA keeps changing and thus is advertised frequently during neighborship establishment until all neighbors enter the Full state. This occupies many system resources and affects neighborship establishment.

H3C allows you to configure OSPF at the (command line interface) CLI to dynamically adjust the LSA generation interval according to the network topology change frequency. The LSA generation interval changes between the specified maximum and minimum interval values. It is incremented by incremental-interval × 2n-2 (n is the number of generation times) each time a generation occurs, up to the maximum interval. When the LSP generation interval has reached the maximum value for three times, it is restored to the minimum interval. With this feature applied in a hub-and-spoke network, the LSA generation interval increases along with the increasing of LSA generation times after the hub device starts to establish neighbor relationships with spoke devices. Thus, the number of packets exchanged during neighbor establishment is reduced. When all neighbor relationships are established, the LSA generation interval restores to the minimum value to ensure high-efficiency routing information exchange.

Controlling the interval for SPF calculation

When the Link State Database (LSDB) changes, OSPF needs to use SFP to re-compute shortest paths. In a hub-and-spoke network comprising a large number of neighbors, frequent SPF calculations consume too many system resources, resulting in packet retransmission and discarding.

H3C allows you to configure OSPF at the CLI to dynamically adjust the SPF calculation interval according to the network topology change frequency. The SPF calculation interval changes between the specified maximum and minimum interval values. It is incremented by incremental-interval × 2n-2 (n is the number of calculation times) each time a calculation occurs, up to the maximum interval. When the SPF calculation internal has reached the maximum value for three times, the calculation interval is restored to the minimum interval. In a hub-and-spoke network, when the SPF calculation times keep increasing during neighborship establishment, the SPF calculation interval keeps going greater to ensure that the hub device can process the OSPF packets sent from the spoke devices in time. When the network becomes stable, the SPF calculation interval restores to the minimum value to ensure high-efficiency update of routing information.

References

·     RFC 2328: OSPF Version 2

·     RFC 4222: Prioritized Treatment of Specific OSPF Version 2 Packets and Congestion Avoidance