Traditional approaches address the issue of reducing carbon footprint in data centers focusing generally on reducing energy consumption of IT equipment and other electronic and networking devices present in data centers.
ECO2Clouds team decided to investigate a different and more innovative approach. We consider not only energy related issues, we address CO2 emissions produced by datacenter infrastructures and by running applications.
ECO2Clouds follows a three-step approach:
Measurement: The first step is to put in place the means for quantifying the environmental impact of cloud computing by using a set of metrics that measure the greenness of running applications and infrastructures on sites. This is done with ECO-Aware metrics defined by the project.
Monitoring and Analysis: Once we have implemented the required set of ECO2Clouds metrics and properly instrumented the Cloud infrastructures with a monitoring system, the ECO2Clouds environment is able to collect and track energy and carbon footprint data from sites, physical and virtual infrastructure, and from applications. All gathered data is then fed into an optimization and deployment model and are analyzed to generate greener application deployment configurations.
Acting Green: With ECO-Accounting in place ECO2Clouds provides different means of enabling greener utilization of modern IT. More concretely, the ECO2Clouds Workload Scheduler technology allows Acting Green to deploy and manage workloads on multi-cloud scenarios. Application self-adaptation possibilities are also considered by using the ECO2Clouds Application Controller.
The ECO2Clouds Solution strives to be a completely agnostic cloud scheduling facility. The core algorithms of ECO2Clouds do not take into account the underlying cloud infrastructure. The ECO2Clouds Solution only assumes that the cloud infrastructure is able to provide the required monitoring data and is also able to start/migrate/stop virtual machines.
Everything works from an agnostic perspective: the definition, assessment, utilization and monitoring of ECO-Aware metrics and the use of innovative workload scheduling and application adaptation strategies included in the components; help optimize CO2 emissions and energy consumption of cloud applications in multi-site cloud infrastructures.
ECO-Aware Metrics: the set of ECO-Aware Metrics is the underpinning pillar that provides the quantification of energy consumption and environmental impact according to ECO2Clouds. All other assets rely on these metrics as the baseline to operate. ECO2Clouds considers the following layers: Infrastructure metrics includes: the host layer; and the site layer; metrics at the virtualization layer characterize the virtual machines on which the applications are running; and application layer metrics capture the behavior, in terms of energy consumption and CO2 emission, of the applications running on the VMs deployed in the infrastructure.
ECO-Aware Accounting: the ECO2Clouds ECO-Aware Accounting is a core asset that is responsible for the monitoring and analyzing ECO-Aware metrics feeds from the underlying cloud and for providing data models the ECO-aware Scheduler and Application Controller. The Accounting component relies on a REST Metrics API Abstraction that provides a generic API for retrieving a constant stream of raw values of metrics from the underlying cloud infrastructure monitoring system used. This mechanism assures that the metrics collection works independently from the monitoring infrastructure used for the cloud.
ECO-Aware Portal: the ECO2Clouds Portal, a user interface intended to act as the entry point for the use and configuration of ECO-Aware Cloud Scheduling System and as interaction channel with the user for reporting and notification of cloud infrastructure utilization and status.
Besides providing ECO-Aware Monitoring and Accounting, two other important components within our stack enable minimizing CO2 emissions and optimizing energy efficiency of cloud applications running on federated cloud infrastructures. ECO2Clouds provides optimization for initial deployment and then for runtime adaptation at virtual machine and application level, as follows:
The ECO-Aware Cloud Workload Scheduler: operates at design-time with the decision about the initial deployment of applications into virtual machines. Users can interact with the portal to manage cloud applications which are then directly sent to the workload scheduler to put in place optimization techniques that reduce the CO2 emissions
The ECO-Aware Cloud Application Controller: enables the means to provide ECO-Aware informed runtime adaptation actions at application aiming to reduce the CO2 emissions exploiting the functionalities provided by the ECO2Clouds Scheduler. It gives more dynamic options for controlling apps behavior and seek for best execution venue under the eco-perspective; and under the possibility to postpone the execution of the application to a better period/location where the energy mix is more convenient (e.g. Application Controller decides to execute an application in Europe at night to benefit from a lower CO2 than in the US).
ECO-Aware Clouds Demo - Design Time
ECO-Aware Clouds Demo - Run Time
Site-level Attribution and Forecasting ov VM CO2