Proposta de uma metodologia Green ICT automatizada Erick Butler Poletto Ricardo Alexandre Fiorelli Prof. Paulo Cugnasca (Brasil) Prof. ssa. Chiara Francalanci (Itália)

Agenda Introdução Objetivos da pesquisa Trabalho na Itália Pesquisa e dados Análise dos resultados Trabalho na Poli Criação de uma metodologia verde Implementação e desenvolvimento Introduction about the topics, green ict and method to apply a green initiative in the corporation, benefits… Present the research objectives and contribution we wanted to achieve with this work… Talk about the methodology we addressed for solving these questions… The final conclusions and contributions Present Further work that can be developed.

Green ICT O que é computação green? Fases do Green ICT Avaliação e Plano de ação Benefícios do Green ICT: Redução de custos Redução de consumo de energia Redução da área ocupada pelo datacenter Estratégia de marketing Assessment: analyze where in the data center the more energy is being wasted; Action Plan: and then to act with correction and prevention interventions. For example: In the last years, the concept of Green ICT has been increasingly popular by the mantra of Going Green. there exists a big gap between what companies think it is a green ICT solution and what they are really doing about it. Focus on the reduction of costs related to energy consumption The costs with power and cooling in data centers can reach up to 20% of the IT cost A Green ICT methodology could be created in order to reduce power and cooling costs

Green ICT: pontos de vista Green do ICT Green pelo ICT Fazer uso de ICT para reduzir o impacto ambiental da organização Reduzir o impacto de ICT no ambiente (our work focus on this one) "Green of ICT" focuses on the actual operation of ICT equipment and information systems. It aims to reduce the burden on the environment through power savings in ICT equipment and the recycling of resources. Green of ICT concentrates on the operation of the ICT equipment and information system. The objective of the method is to reduce the environmental impact through power saving and recycling. Data Center Centralizing the information in a Data Center, with consolidated data, thereby reducing power consumption while improving the efficiency regarding cost, operation and maintenance. Hosting It is an enterprise related solution including virtualization, which can be used to optimally allocate resources. "Green by ICT" focuses on the efficiency improvements brought about by the active use of ICT. It aims to reduce logistics and human movement through the sharing of information utilizing ICT. Green by ICT concentrates on the energy efficiency reached by the use of ICT operations. The objective is to reduce logistics costs and personnel transport through ICT. Communitation reduces the energy consumed with transportation, by utilizing VOIP solutions with video-conference Remote Data Access reduction of travel and material consumption by having, for example, mobile terminals or VPNs, where the employees can have access to work information wherever. In addition, the service should provide secure and reliable remote access enabling the use of multiple devices.

Objetivos da pesquisa Criação do banco de dados de componentes Avaliar uma ferramenta de benchmark, em relação ao consumo energético (SANDRA) Criação de uma metodologia verde Identificação de pontos críticos de consumo em uma empresa Obtenção de dados dos componentes Auxilio na implantação de uma política verde Objectives Create a database of components, in order to help in the Assessment phase of the green solution Explain what is the Database of Components: (how we build we are going o address it later on methodology) The main source of information used in the database was a benchmarking tool called SiSoftware Sandra Sandra provides information about computer components regarding performance, pricing and power Evaluate the validity of a benchmarking tool (SANDRA), in order to help the development of a green solution Explain what is SANDRA Benchmarking tool, with a set of It contains a vast database of components associated with both benchmark results and manufacturer specifications. Benefits Objetivos Benefícios

Pesquisa sobre o estado da arte Configuração das máquinas Políticas / Ferramentas / Labels Arquiteturas Thin Client Virtualização dos Servidores Armazanemanto de Dados Arquiteturas de distribuição de energia Sistemas de Refrigeração Explain what is each of these categories, one sentence each. Machine Configuration Processor, RAM (little part 2W, 3Wmax), Hard-drive, Chassis, Monitor Policies / Tools / Labels Thin Client Architectures Virtualization in servers Data Storage Power Architectures Cooling

Sandra (ferramenta benchmarking)

WebSPHINX (web crawler) In this slide we present the method with which we built the database…

Instrumento de medição Validação dos dados do SANDRA

Análise dos resultados: Medição Comparação entre: Os dados estimados pelo SAMNDRA Consumo nominal do processador Potencia medida pelo SANRA > potencia nominal 30% dos componentes

Análise dos resultados: Comparação Processador inativo e fully stresssed Comparison between 3 results: Sandra estimated processor/chipset/memory consumption Measurement made with idle computer – estimated PSU power Measurement made with stressed computer – estimated PSU power Results: Average difference of 39% for the idle computer measurements 23% for the stressed computer measurements

Análise dos resultados Dados não confiáveis do SANDRA? As estimativas do SANDRA estão maiores que a potência nominal As medidas diferem, por uma larga margem das feitas com o instrumento e medição Sandra provides higher component power estimation than the component nominal power in some cases Sandra power estimation does not correspond by a large margin to the measurements made with the measurement tool

Análise dos resultados Possíveis razões do dados observados: Não contralização dos dados Sandra pode associar o componente a outro em seu banco de dados Informação da potência está errada Potência não associada ao consumo do computador Refere-se ao valor inativo ou fully-stresses Erro na fonte das medições: HD e wireless << Processador, memória ou PSU Oscilações nas medições manuais Uso de estimativa do monitor e da PSU Use of PSU and monitor power estimates As the components present in the benchmarks are not uniquely identified by a MPN code, the searches are made with the name of the component. This could associate the benchmarked component to a different component in Sandra’s database The power-related information could be itself erroneous The power-related information is not associated with a given machine workload

Outline Trabalho na Itália Trabalho na Poli Introdução Objetivos do trabalho na Itália Metodologia Análises e conclusões Trabalho futuro Trabalho na Poli Implementação da base de componentes Desenvolvimento e implementação de uma metodologia Green Conclusão Introduction about the topics, green ict and method to apply a green initiative in the corporation, benefits… Present the research objectives and contribution we wanted to achieve with this work… Talk about the methodology we addressed for solving these questions… The final conclusions and contributions Present Further work that can be developed.

Conclusions The database of components is a very effective tool With relation to power consumption, Sandra is not the best source of data Other sources of power-related data should be considered Although the software estimates were diverse with respect to the direct measurements, the usefulness of the component database should be no lesser. It provides a great number of other useful information, mainly regarding component performance benchmarks and price which are essential when analyzing datacenter equipment. Another possibility is to use other sources of power-related data other than the Sandra benchmarking tools, such as systematic measurement of components in an adequate aggregation level or even to define a function of the components nominal power which would approximate their power consumption at a given workload