ECO4CLOUD-HP PROOF OF CONCEPT RESULTS Eco4Cloud is a spin-off from the Institute for High-Performance Computing and Networking of Italy’s National Research Council (CNR) which has researched/developed/engineered an innovative and very effective solution to optimize the consolidation of virtual machines and cut 30-60% (measured!) the energy bill of highly-virtualized/large-scale data centers. Furthermore, in typical scenarios our Value-Pricing business model returns pay-back times <<1 yr and 3yrs ROI >120%. How does it work? In a nutshell, it’s an innovative bio-inspired probabilistic algorithm which consolidates the max number of virtual machines on the min number of physical servers in a data center, enabling the switch-off/hibernation of those freed-up, making them dynamically available as additional capacity for incremental workloads. The value proposition is compelling and crucially relevant for both business as well as technical audiences, and goes far beyond energy saving alone: monitoring, capacity planning, orchestration optimization, risk monitoring/mitigation. This document is a summary of extensive and rigorous Eco4Cloud tests that took place at HP’s Innovation Center. ECO4CLOUD-HP PROOF OF CONCEPT RESULTS REDUCTION OF DATACENTER ENERGY BILL In a virtualized datacenter on average only 20-30% of server capacity is utilized. Still, an idle server consumes 6570% of the power consumed when it is fully utilized. By consolidating the maximum number of Virtual Machines (VM’s) on the minimum number of physical servers, the non-utilized servers are set to hibernate, hence eliminating their energy consumption and reducing up to 60% the overall energy bill. Furthermore, by having less energy consumed by the physical servers, power cooling and sizing gets reduced as well, contributing to indirect additional energy (OpEx) and CapEx reductions respectively. CONSOLIDATION AND EFFICIENCY The solutions/techniques commonly available today are semi-manual, extremely complex, poorly adaptive and not scalable. The VM’s-consolidation problem requires exponential time if solved with a typical deterministic approach, which translates into sub-optimal consolidation and efficiency figures. E4C innovative statistical algorithm and distributed software architecture helps achieve figures in eccess of 90% CAPACITY PLANNING Thanks to the optimal occupancy of physical resources and adaptive optimization of inherently variable workloads provided by E4C, CIO’s can improve the accuracy of the budget and capacity planning for their Datacenters, hence optimizing their CapEx and OpEx. MEET DC SLA’S - REBILIABILITY, AVAILABILITY, PERFORMANCE Thanks to the insights and real-time monitoring analytics of critical system parameters (e.g. CPU load, memory usage, bandwidth saturation, temperature, storage occupancy) provided by E4C, Datacenter Managers can proactively/predictively prevent SLA violations. E.g. E4C intructs the move of VM’s from servers before they get overloaded. Furthermore, threshold-triggered dynamic responses via system-alerts and automated tasks help mitigate risks and overal datacenter reliability. EASE OF DEPLOYMENT & MANAGEMENT – HYPERVISOR AGNOSTIC E4C is extremely easy to customize and offers the option for virtually unlimited customized rules to allocate and migrate VM’s. Intuitive graphical reports on utilization of servers, workload distribution and power consumption simplify operations. E4C does not interfere with the management of the virtualization platform(s), but rather is an add-on that interoperates with existing DCIM platforms and hypervisors making the datacenter more efficient. ECO4CLOUD-HP PROOF OF CONCEPT RESULTS Furthermore, E4C is available on multiple virtualization platforms (e.g. Vmware VSphere, Microsoft Hyper-V, KVM), hence ideal for hybrid environments. SCALABILITY Thanks to its adaptive/self-organized distributed algorithm, E4C is very efficient for such a complex (xN) problem, and converges to optimal state regardless of DC size in a just a few hours. This is a great advantage as energy and cost savings are proportional to the datacenter size. ECO4CLOUD-HP PROOF OF CONCEPT RESULTS HP and Eco4Cloud ran a deployment that demonstrated how energy-saving features provided by Eco4Cloud, running on the latest generation of HP x86 servers, can help customers significantly reduce energy consumption in the Data Center, with no impacts on business services. Eco4Cloud can improve the performance of a data center in a typical scenario where resources are under-utilized. The Eco4Cloud software can greatly improve efficiency of servers (up to 80-85%) by consolidating the VMs on a lower number of servers and putting some servers in a low power state. Moreover, Eco4Cloud prevents any possible event of server overload, therefore energy and cost savings are obtained without deteriorating the perceived Quality of Service. The results on the deployment performed on HP servers show that Eco4Cloud effectively consolidates the workload, and yields a reduction of power consumption ranging between 40% and 50%, depending on the chosen configuration of Eco4Cloud. Assets energy consumption becomes proportional to the resources utilization, and not constant, as without using Eco4Cloud, regardless of the servers utilization. An estimation of consumed energy, based on the deployment performed on the HP servers, shows that the energy savings are equal to about 48.1% with respect to the prior situation, in which Eco4Cloud was not deployed.