NAD 3020 manual High Availability, Design Goals, Design Guide

Design Guide

Layer 2 adjacency in the server farm allows for the deployment of servers or clusters that require the exchange of information done at Layer 2 only. It also readily supports access to network services in the aggregation layer such as load balancers and firewalls, enabling an efficient use of shared, centralized network services by the server farms. In contrast, if services are deployed at each access switch, the benefit of those services is limited to the servers directly attached to the switch. It is easier to insert new servers into the access layer when the aggregation layer is responsible for data center services, and the Layer 2 environment provides the flexibility to scale the number of ports -- another benefit provided in a Layer 2 access layer.

The access layer must provide a deterministic environment to help ensure a stable Layer 2 domain. A predictable access layer allows the spanning tree to converge and recover quickly during failover and fallback scenarios.

High Availability

High availability in the data center is a goal that must be achieved systematically. A highly available environment is attainable by addressing each layer of the data center and each of the devices that comprise that particular data center layer. Network and software features help achieve high availability, as well as physical redundancy of links and devices.

The aggregation and access layers use redundant devices and links to help ensure no single point of failure occurs. The Layer 2 and Layer 3 features supported by these switches also create a highly available infrastructure. Spanning Tree Protocol support on both the aggregation and access switches creates a deterministic topology that converges quickly. Logical redundancy or fault tolerance may be achieved with Layer 3 technologies such as Hot Standby Router Protocol (HSRP) or Virtual Router Redundancy Protocol (VRRP). These protocols allow for virtualization of the gateways for servers or clients across the physical routing devices in the network. This virtualization mitigates the effect of a routing-device failure on the availability of data center services. Load-balancing services deployed in the aggregation layer allow the network to monitor server health and application availability. These devices and features combined produce a more resilient application environment.

Dual homing a server in relation to separate access layer switches is another method to achieve a higher level of availability in the data center. NIC teaming removes the possibility of a single NIC failure isolating the server. It requires the server to have two separate NICs that support teaming software. Typically, teaming software detects failures over an external network probe between members of the team by monitoring the local status of each NIC in the team. The combination of dual-homed servers and a network load balancer provides an even greater level of availability for the server and the applications it supports.

Design Goals

This section describes the design goals for deploying blade servers and the functions that the Cisco Catalyst Blade Switch 3020 supports in data centers. It discusses high availability, scalability, and management.

High Availability

Data centers are the repository of critical business applications that support the continual operation of an enterprise. These applications must be accessible throughout the working day during peak times, and some on a 24-hour basis. The infrastructure of the data center, network devices, and servers must address these diverse requirements. The network infrastructure provides device and

© 2008 Cisco Systems, Inc. All rights reserved. This document is Cisco Public Information. Page 12 of 28