Periodic Reporting for period 2 - EcoMultiCloud (HIERARCHICAL APPROACH FOR GREEN WORKLOAD MANAGEMENT IN HETEROGENEOUS AND INTERCONNECTED DATA CENTERS)
Reporting period: 2016-12-01 to 2017-11-30
#1: Complete development and integration of EcoMultiCloud modules and infrastructure;
#2: Assessment of EcoMultiCloud effectiveness and final validation in a real case pilot;
#3: Successful final business innovation plan and pre-commercialization activities."
Starting from the data available at Eco4Cloud regarding the TIM infrastructure, HW resource metrics like CPU, RAM and Disk usage were listed and taken into account, and performance indexes (CPU ready time, ballooned memory and active memory) investigated. Migrations among the TIM data centers were detailed, and collected data was processed in order to demonstrate the impact of migrations in the whole system. The obtained results showed that the quality of the service perceived by the user is high, thus confirming the validity of our approach regarding remote migrations.
The algorithms were customized according to the features of each selected DC owned by TIM (located in Rozzano, DC1, Pomezia, DC2 and Bari, DC3). Collection of the results on the testbed performance involved the parameters on carbon emissions for the three DCs, the energy cost, the consumed energy, the distribution of the workload among the DCs and the number of migrations performed. The alpha version of the EcoMultiCloud software was successfully released.
The final pilot tests of the product were planned and conducted concretely. The three DCs, initially considered as equal in architecture, number and hosts capacity, were characterized with their real features, the assignment function driving the workload distribution among the different DCs was customized, and the pilot tests were performed, first in a passive mode and then in an active mode; performance data was collected for the assessment of the tests results. The beta version of the software, updated and improved with the support of TIM staff, was released. Constraints on workload mobility were taken into account in the final version of the SW, the algorithm further refined and the GUI appearance and usability improved. Finally, resources have been devoted to the implementation of commercial and communication activities; a dedicated dissemination plan was implemented supporting the diffusion of the EcoMultiCloud project results; a final business plan was prepared, and financial forecasts were updated with the most recent estimates on production costs and product prices.
- Integration with other DC management tools - EcoMultiCloud architecture can be easily integrated with facility-side DC Infrastructure Management tools.
- Availability on multiple virtualization platform – EcoMultiCloud works on top of any virtualization platform, ideal for hybrid environments;
- The amount of inter-DC bandwidth available could be a limiting factor – The adaptive and self-organizing nature of our algorithm allows to handle the problem of the amount of inter-DC bandwidth available.
- The challenges posed by increased workload portability among distributed DCs- our solution envisions that all DCMs in a group of EcoMultiCloud-enabled facilities will communicate with each other.
The current year is going to be the first profitable fiscal year for Eco4Cloud. The profit and loss account forecast of the company estimates a sales revenue growth for the company of 200% year on year, thanks to the features introduced in the Eco4Cloud products suite after completing the EcoMultiCloud project, financed by the H2020 SME INST. Using EcoMultiCloud in the private cloud enables companies to achieve their SLAs at the lowest cost, mainly through workload migration and applications stacking. In a hybrid cloud environment, EcoMultiCloud helps with dynamic placement decisions according to policy, removing people from the provisioning/brokering process. It also enforces policies that reduce sprawl and wasted costs, increases policy compliance, and reduces risks from accidental or malicious activities. Dynamic optimization technology for the public cloud enables better utilization of services at lower costs. The feature of multiple clouds management, provided by the EcoMultiCloud project, allows the company to make its products suite more appealing for companies with large IT infrastructures geographically dispersed, and therefore will allow to increase sales, that will exceed 1M€ in 2019.