IF I DEPLOY ECO4CLOUD, WHAT ELSE DO I NEED TO MANAGE MY DC?
Eco4Cloud is a platform that operates on top of most popular Server Virtualization infrastructures, e.g. VMWare vSphere or Microsoft Hyper-V. Therefore, Eco4Cloud doesn’t require anything other than the virtualization software a data center is already running and actually integrates and interacts with it.
HOW DOES ECO4CLOUD INTEGRATE WITH THE SERVER VIRTUALIZATION PLATFORM?
Eco4Cloud relies on the libraries offered by each virtualization platform to retrieve the information needed by the Eco4Cloud algorithm, e.g. CPU and RAM utilization on local servers. All Eco4Cloud operations (e.g. a VM migration or a server switch) are passed to and checked by the virtualization platform before being executed. Therefore, Eco4Cloud operations do not conflict in any possible way with the normal operation of the underlying virtualization platform but rather feeds it.
IS THIS LIKE LOAD BALANCING?
Load balancing algorithms tend to spread the workload across a number of servers, and prevent situations in which some servers are underutilized and others overloaded. Load balancing does not aim to unload and switch off servers, in fact energy saving is not the purpose of such algorithms.
Eco4Cloud consolidates the applications (hosted by Virtual Machines) onto as few servers as possible to reduce costs and energy consumption, and increase utilization. Since Eco4Cloud can power-off or hibernate servers which are not required to meet the services demand, this translates into a reduction of the electrical energy consumed for both computation and cooling. So, while load balancing distributes the load over all the available servers which may lead (and generally leads) to underutilization of all servers, consolidation clusters the load on as few servers as possible. As a result, with VM consolidation (which Eco4Cloud features) physical servers are either switched-off or used efficiently, with resources (e.g. CPU, RAM) operating within a predetermined range of utilization (or thresholds), e.g. between 50% and 85%.
In addition to cost reduction and energy savings, workloads consolidation brings other important advantages, such as:
- carbon emissions are significantly reduced, which can help companies qualify for “green certifications”;
- powered-off servers can be used to accommodate additional workloads, hence deferring the purchase and related CapEx of additional equipment/capacity;
- given a certain workload, fewer servers will need to be managed/maintained/cooled;
- real estate (i.e. data center square footage) requirements also get reduced as fewer physical servers are in operation.
WHAT OTHER TOOLS CAN I REPLACE IF I DEPLOY ECO4CLOUD (SPECIFY BRANDS)?
By default, server virtualization software does not perform optimized consolidation of VMs and therefore you do not need to replace anything. Eco4Cloud is in fact an add-on element which makes your data center more efficient and your physical servers utilized to their full capacity. Among commercially available server virtualization platforms, only VMWare offers a module which partially addresses VM consolidation, i.e. the Distributed Power Management (DPM) tool. However, the algorithm used by DPM is not scalable and does not allow the user to control and monitor the distribution of the workload, which instead Eco4Cloud features. It is our evidence that VMware users generally prefer not to activate DPM at all because its behavior is not clear and it is not controllable/configurable.
HOW DO I KNOW ECO4CLOUD WILL NOT IGNITE UNDESIRED BEHAVIORS IN MY DC OPERATIONS?
Robust measures are designed into Eco4Cloud to ensure the safety of your data center. Eco4Cloud in fact prevents/anticipates potential overload conditions which may have an impact on the quality of service experienced by users and on your data center’s SLAs. This is achieved by setting safeguard thresholds on server usage. This not only safeguards servers from being overloaded and therefore provides an inherently safe solution but, due to the self-organizing and dynamically automated allocation of VMs, significantly reduces the burden related to the administration and monitoring of the data center.
MAY ECO4CLOUD LOWER MY DC SECURITY?
Not at all.
In fact Eco4Cloud shares/inherits the same security restrictions and attributes (e.g. access and authorization policies, firewall rules etc.) as the underlying server virtualization platform. This is because Eco4Cloud itself is configured as a Virtual Machine that runs on a specific server of your data center (typically the server used as the manager of the virtualization platform).
HOW LONG WILL IT TAKE TO FULLY DEPLOY ECO4CLOUD?
Eco4Cloud can be set-up and running in 2 weeks.
More specifically, a typical Eco4Cloud deployment articulates in the following stages:
- Analysis – Eco4Cloud uses a monitoring tool to analyse the data center (1 week);
- Simulation – Eco4Cloud estimates the impact of virtual machine consolidation in terms of energy savings and QoS improvement (3 days);
- Installation and early monitoring – Eco4Cloud is installed on the data center, full on-site support is provided(3 days);
- Remote monitoring – The data center is constantly monitored and product upgrades are provided.
WILL I EXPERIENCE ANY DOWNTIME DURING INSTALLATION?
Eco4Cloud starts optimizing as soon as it is installed but this does not affect the operation of your existing services in any possible way.
WHAT HAPPENS IF ECO4CLOUD STOPS WORKING OR HANGS-UP?
In case of such an occurrence, an alert is immediately sent to the data center administrators through a “watch dog” mechanism. Anyway the only implication of having Eco4Cloud not running is that energy and cost savings can gradually revert to the sub-optimal levels where they were before its introduction.
WHAT OTHER COMMERCIAL PRODUCTS PROVIDE "VM CONSOLIDATION"?
There are a few products on the market which address this space, however due to their traditional approach the savings they are able to achieve are only a fraction of what can be achieved by Eco4Cloud; furthermore, administration is much heavier and complicated, and scalability is not guaranteed (i.e., they are efficient only in small data centers deployments).