Data Center Network Enhancement & Migration

Design Objective

The design objective of the proposed network is to provide a robust, reliable, secure and scalable network infrastructure to customer to support upcoming business requirement. The following are considered into the design:

Provide a consolidated backbone infrastructure amongst three of sites, namely Site1, Site2 and Site3
Provide a flexible solution that is highly scalable on both interface ports and bandwidth
Provide a seamless migration solution to minimize the impact on customer operation.

Ensure existing services can be provided by the new network and additionally new services can be added without major interruption.

Existing Services

Single homed direct static route customers
Single homed BGP customer
Dual homed VRRP customers
Dual homed BGP customer
Multi-floor single homed customer
Multi-floor dual homed customer
Multi-site single homed customer
Multi-site dual homed customer

Existing Network Review

The existing network consists of two Hub sites and one Spoke site, comprising of both Juniper Networks products as well as Cisco products. We have identified several deficiencies with the current network architecture:

Spanning Tree Protocol Issues

The Spanning Tree Protocol has been deployed to prevent looping and provide network redundancy. At the meantime, however, STP brings in the flapping problems that may occur in the network to cause traffic ports switch between UP/Down abnormally.

BGP routes limitation

Current network can only support BGP routes in a limited capacity. E.g. local IPv4 routes only. Due to limited IPv4 BGP routes, default route is required in current network, which causing inefficiency routing process. We would propose to use Juniper equipment to adapt IPv4 and IPv6 full routes, enhancing routing performance and prepare for coming IPv6 network.

Network Capacity and Scalability

The 10G interface is not available on current Cisco Catalyst 6509 switch of this network. What is more, several proprietary protocols such as ISL, PVST+ implemented on the network may create compatibility problems due to non-standardization. All these issues could greatly hinder your valued network from scaling and maintaining capacity to catch up with the industry requirements.

Inter-data center connectivity

Currently, customer is using Layer2 as inter-data center connections, which may have introduce several limitations in term of services, failover time, traffic engineering.

Network Management

A total of eight network cores have to be managed in the existing network, which may greatly raise the complexity to synchronize the network configuration and to keep management accuracy, thus lowering the management efficiency.

We understand that customer will upgrade your existing access layer to eliminate to use of spanning tree, increase system capacity and ease manageability, therefore in our proposed solution, we will be responsible for the core layer and the edge layer. After the migration, it is expected that the aggregation layer will be consolidated with the core layer and the following 10G port per core node are made.

Proposed Network Architecture

There are three sites in the proposed network, namely Site1, Site2 and Site3 site. Each site consists of the edge, core and access layer. The access layer comprise of Cisco access switch, which will be stacked and will have two uplinks towards the core layer. At floors A,B,C at Site1 there are a single switch stack. On the floor D there is three switch stacks, making a total of five switch stacks at Site1 access layer towards the core layer.

Similarly, for Site2, the information given is that there will be five stacks of access layer towards the core layer, as depicted in the diagram below:

At the Site3 site there is a single stack switch with two core routers, and no edge router, therefore the

Maintaining the existing number of inter-site connections, the overall architecture: