Book Image

Learning OpenStack Networking (Neutron)

By : James Denton
Book Image

Learning OpenStack Networking (Neutron)

By: James Denton

Overview of this book

OpenStack Neutron is an OpenStack component that provides networking as a service for other OpenStack services to architect networks and create virtual machines through its API. This API lets you define network connectivity in order to leverage network capabilities to cloud deployments. Through this practical book, you will build a strong foundational knowledge of Neutron, and will architect and build an OpenStack cloud using advanced networking features. We start with an introduction to OpenStack Neutron and its various components, including virtual switching, routing, FWaaS, VPNaaS, and LBaaS. You’ll also get hands-on by installing OpenStack and Neutron and its components, and use agents and plugins to orchestrate network connectivity and build a virtual switching infrastructure. Moving on, you’ll get to grips with the HA routing capabilities utilizing VRRP and distributed virtual routers in Neutron. You’ll also discover load balancing fundamentals, including the difference between nodes, pools, pool members, and virtual IPs. You’ll discover the purpose of security groups and learn how to apply the security concept to your cloud/tenant/instance. Finally, you' ll configure virtual private networks that will allow you to avoid the use of SNAT and floating IPs when connecting to remote networks.
Table of Contents (21 chapters)
Learning OpenStack Networking (Neutron) Second Edition
About the Author
About the Reviewers

Integrating load balancers into the network

When using the HAProxy driver, load balancers are implemented in one-arm mode. In the one-arm mode, a load balancer is not in the path of normal traffic to the pool members. The load balancer has a single interface for ingress and egress traffic to and from clients and pool members.

A logical diagram of a load balancer in one-arm mode can be seen in Figure 10.1:

Figure 10.1

In Figure 10.1, a load balancer is configured in one-arm mode and resides in the same subnet as the servers that it is balancing traffic to.

As a load balancer in one-arm mode is not the gateway for the pool members that it is sending traffic to, it must rely on the use of source NAT, or SNAT, to ensure that return traffic from the members to the client is sent back through the load balancer. An example of the traffic flow can be seen in Figure 10.2:

Figure 10.2

In Figure 10.2, the load balancer receives a request from the client and forwards it to WEB1. The load balancer then modifies...