Recently, we recommended and implemented a transformative energy efficiency plan following these strategies and technologies for a Fortune 500 company:
- Identify optimal location for central data center – The customer’s main campus was selected due to its proximity to the Company’s combined heat and power (“CHP”) system, the proximity to the facility’s very efficient chilled water plant (“CWP”) (with back-up cooling), the proximity to its headquarters, and the existence of one of the company’s largest conventional data centers at the location.
- Upgrade physical servers to high performance state of the art technology – In order to run virtualization/loadshare platforms and to provide satisfactory response times under an Infrastructure-as-a-Service (IaaS) environment, very powerful blade servers must be purchased. These new servers will be quad core/dual processor, 2.5 to 3.0 GHz and be specifically designed to support the virtualization of the hardware platform layer running multiple simultaneous applications or jobs distributed over clusters of physical servers. Unlike the previously installed blade server technology, the new servers are designed to serve multiple, simultaneous user loads or application environments and operate at average CPU utilization levels of 50% or more.
- Implementation of virtualization layers into the data center – Installation of virtualization layers will enable IT management to separate the physical configuration of the servers from a virtual or “apparent” server configuration as required by the various users throughout the company. This means IT management can optimize physical data center asset operation while retaining the flexibility needed to meet the needs of the user base. Virtualization software enables near-dynamic provisioning of compute capacities through abstraction of physical hardware, combining the aggregated demand of all users to run across multiple consolidated physical servers.
(This is achieved by enabling a cluster of servers to support multiple simultaneous applications and processes. The result is that fewer physical servers are required to support a given number of applications and users. The physical servers in the cluster actually share and allocate their compute capacity based on compute demand, thereby increasing the number of applications and users per server. This sharing capability enables the servers to operate at 50% or higher CPU utilization compared to 5% – 10%, which is their historic utilization rate for servers.)
- Migrate to containerized data center support structure – The increase in compute capacities of the physical servers/utilization by Infrastructure-as-a-Service (IaaS) create higher concentrations of heat per given unit of rack space in the data center. To be able to remove this heat and transfer it out of the data center space will require an extremely efficient transfer environment. The current conventional data center at the customer cannot effectively achieve this.
(The customer, therefore, plans migrate to a containerized platform for storage and operation of the high performance servers. The container is a self-contained enclosure specifically designed to circulate air to transfer heat from the physical servers to heat exchangers which transfer the heat into the chilled water loop. The containerized storage approach uses about 10% of the amount of energy required by conventional data centers to transfer this heat. The container achieves these efficiencies through virtual elimination of hot/cold air mixing, CRAC conflicts, air turbulence and air travel distances.)
- Decommission/limit growth of disparate data center operations – Once the container, servers and virtualization platform is installed, the IT group will begin the process of migrating applications and user load from the company’s existing disparate and inefficient physical platforms to the new efficient one. This will be achieved by establishing a virtual replica of the old data center and migrating the applications onto it thus achieving total transparency for the user base. Once the migration is complete, the old data center will be decommissioned.
The execution of these steps resulted in a rapid increase in the utilization of the new data center. Significant annual energy savings will be achieved also. The final result will be an approved application that will authorize payment of incentives once the project is complete and the savings are verified.