DESIGN ANALYSIS
HP's redesigned network incorporates switch lines that contain Layer 2 or 3 switches in configurations and port densities to fit neatly into any deployment scenario, from the access layer--where wiring closet issues such as power, cooling, and hands-off management are as important as port counts--to core switches that require high availability, high performance, and high port counts.
Starting at the core, HP replaced TacDoh's redundant data center switches and core switch-router with two ProCurve 5406zl-48G Intelligent Edge switches with redundant power supplies. A premium license is an additional $2,999 per unit, but it's required for the Virtual Routing Redundancy Protocol, which provides failover for routing. HP's reasoning for collapsing the data center servers and core switch router onto two 5406-48G switches is to simplify the network topology and provide an aggregation point between the data center and the access switches.
The 5406-48G is a six-slot chassis that ships with 48 10/100/1,000 Ethernet ports, which take up two slots and a single power supply. HP added an additional power supply for redundancy. There are four slots available for growth. Given the number of ports required for the data center servers and the access switches, TacDoh will have plenty of ports available as its network needs grow.
HP's redesign uses switch chassis from the 5400zl line--one for each department--populated with 10/100/1,000 Power over Ethernet ports to support the existing user base. Two smaller switches from the 3500 line are used for locations that have fewer ports. The access switches are underpopulated in the design; TacDoh plans to double its workforce in the next year and the chassis will allow incremental expansion.
HP chose to use chassis versus individual stacking switches to improve performance.
HP's solution has a number of high-availability features. The access switches are dual-homed using 1000Base-T uplinks to both core switches and Layer 2 protocols such as the Multiple Spanning Tree Protocol and the Link Aggregation Control Protocol. The zl modules are hot-swappable and and can be shared among the 5400zl chassis, so fewer spares need to be kept on hand. In addition, modules, mini-gigabit interface converters, and power supplies are hot-swappable.
We also wanted to make sure that the chosen vendor's products will support changes as the network expands, both in size and in the types of applications supported. HP's infrastructure support goes beyond speeds and feeds. Traffic monitoring uses industry standard protocols such as RMON for remote monitoring and sFlow for monitoring network traffic flows, and provides integration with third-party monitoring tools. However, sFlow is generally less supported than Cisco's NetFlow, so your choices of third-party applications that can use flow data may be limited.
Management is through HP's ProCurve Manager Plus management tool. The tool, known as PCM+, offers a full suite of management tools for device discovery, configuration management, monitoring, troubleshooting, and Layer 2 and Layer 3 mapping. Designed for small installations, PCM+ provides a wealth of tools to manage the installation. HP's command line interface has useful features as well. CLI policies can automate repeated configuration tasks and can be automatically executed by an event within the switch.
