Case Study (Headquarters)

LAN/WAN  Architecture

New York City Headquarters


The LAN/WAN infrastructure topology that will be deployed in XYZ's HQ is shown in the diagram above. This network was staged in a lab environment. There are a few design considerations, which we believe should be considered to help improve the following:

·       Ability to troubleshoot the design

·       Performance

·       Redundancy

·       Ability to scale to meet future requirements 

However these design considerations, listed below, are not critical for the design to function and XYZ  has opted to not implement these recommendations due to economic reasons:

·       A full mesh at the distribution level will provide added stability and accomplish XYZ's request that local traffic not traverse the FLEXWAN switches.

·       If a distribution level full mesh is implemented then the position of digital gateway is optimal, otherwise our recommendation is to "hang" the digital gateways off the FLEXWAN switches.

·       A layer three connection is recommended between the two FLEXWAN switches so that traffic traveling from one remote office to another and/or other XYZ WAN traffic is not required to traverse the distribution switches at Corporate HQ.

·       Implementation of Etherchannel uplinks to the voice gateways is not recommended with this design.

·       The server farm will be connected to the distribution switches with no L2 links between them.  Ideally, it is Cisco’s recommendation that a server farm be created with independent and separate distribution switches for fault isolation and protection from L2 loops resulting from DUAL NIC connectivity.

·      Currently, XYZ's WAN backup strategy consists of only four T1s, which may not be sufficient to support voice and data traffic in the event of a DS3 failure.  Based on this design, a manual reconfiguration of the CallManager Servers will be required to recover from a telephony outage caused by a DS3 WAN failure for medium branch office (BR1111).  This design trade off may not be practical due to the time requirement to reconfigure the network to implement voice compression, and transcoding resources.