The Dymon corporate infrastructure was operating on unique physical hardware per server and per service.  We identified 8 physical servers of identical capacity that were running at less than 20% utilization, in one instance 0% utilization.  Switching and networks were analyzed and we identified 3 totally unused Layer 3 switches with zero utilization.

A plan was put forward to migrate to a fully virtualized infrastructure, segregate corporate traffic with various VLANs to support operations, POS systems, publicly accessible web services, CCTV data, and credit transactions in addition to iSCSI traffic, virtualization traffic, monitoring traffic, and Building Automation traffic.

Primary objective was to re-architect the corporate network to support separation of data based on type with QoS and rate limiting, implement a robust and fault tolerant virtualization stack to eliminate all single points of failure and utilize un-allocated equipment to enable growth and enhance redundancy.

Secondary objectives were to streamline the requirements for operation and eliminate frequent downtime occurrences, whilst improving the efficiency and efficacy of the network as a whole, and enhancing backup procedures.

The primary challenge was soft challenge realized as a distrust and fear of virtualization technologies.  We mitigated this with extensive training and exposure to real-world examples of implementation with our test lab.  Key focus was placed on demonstrating the benefits of virtualization with physical failures failing over within the cluster, speed to create and implement VMs, and the benefits of features such as snapshot technologies.

Our secondary challenge was working with a zero-value budget.  Physical servers were re-purposed by creating a rolling migration plan to migrate a physical system and then convert the hardware into a hypervisor.  Xenserver was selected as the hypervisor of choice due to its zero cost whilst supporting HA clustering, with the option for vendor support from Citrix in the future as required.  Various QNAP SANs were being utilized for CCTV storage, and these were re-purposed into a “poor mans” cluster with simplistic pooling and data replication handled either by the SANs or via VMs.