Cisco Cisco Prime Network Services Controller Adaptor for DFA 백서

 

 

© 2016 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public Information. 

Page 3 of 32 

Introduction 

The primary goal of this document is to provide guidelines about how to implement application load balancers in 

the data center using Cisco

®

 DFA (Dynamic Fabric Automation). 

Readers will learn how to integrate load balancers into the DFA Fabric using network autoconfiguration on Cisco 

Nexus

®

 Family switches. The network integration deployment scenarios covered in this document are not specific 

to any vendor and can accommodate any application load balancer available on the market today. 

This document is written for network architects; network design, planning, and implementation teams; and 

application services and maintenance teams. 

This document assumes that the reader is already familiar with the mechanisms of the DFA autoconfiguration 

feature. The reader should be familiar with mobility domain, virtual switch interface (VSI) Discovery and 

Configuration Protocol (VDP), network profile, and services-network profile configurations. Please refer to the 

following configuration guide for more information: 

. 

Placing the Application Load Balancer in the Fabric 

Load-balancer appliances can be connected in several places in the network. 

Network autoconfiguration on Cisco Nexus switches allows dynamic instantiation of the necessary configuration on 

leaf nodes, so the recommended approach is to connect load balancers at the leaf level. Spine nodes do not 

contain any classical ethernet (CE) host ports and should not be used as service attachment points.  

With the dynamic autoconfiguration feature, load balancers, in both hardware and virtual machine form factors, can 

be connected anywhere in the network. Network utilization and forwarding can be optimized when relevant service 

appliances are attached to a single pair of leaf nodes, referred to as the service leaf. The logical role of the service 

leaf does not change the configuration or enable additional features on this set of leaf nodes. It is used essentially 

as a central location for attaching service nodes. 

If your organization chooses to use the service leaf and needs to use virtual load balancers or virtual appliances, 

you will need to follow certain guidelines. With automated or orchestrated virtual services deployment mechanisms, 

the automation or orchestration tool must help ensure the location of deployed virtual services and virtual 

machines. For example, in Cisco UCS

®

 Director, you can specify a set of hypervisors, on which virtual services can 

be created. Attaching this set of hypervisors to the service leaf will help ensure the location of deployed services in 

the network. 

In a network, a load balancer can be deployed in the following scenarios: 

● 

One or more load balancers for a given tenant: Load balancers can be virtual or physical. 

● 

One or more load balancers shared across multiple tenants: Here, the load balancer is most likely a 

hardware platform, and depending on the vendor and software, the load balancer may provide built-in 

virtualization features, such as traffic domains, Virtual Routing and Forwarding (VRF) functions, and virtual 

contexts.