Vitenskapelig sammendrag
This project is in the problem domain of system administration. As the size of deployements and datacenters increase, the complexity of systems and their constellation grows beyond what can be managed by traditional system administration approaches and current tools. The onus falls on the local administrator to overcome the challenges locally and on their own. More work is needed in order to understand and generalize they challenges and approaches in order to move the profession and its practices forward. Of particular intrerest are cloud technologies, which have the promise of delivering increased flexibility and effectiveness through decoupling the system from the hardware and offering self-service functionality to its users. Needless to say, viurtalization and cloud computing are a great enabler, but at what cost?Today's clouds are still under heavy development, meaning they change often and documentation is sparse. For a system administrator it is a huge task to properly set up and manage a cloud environment. This is not possible in an straight-forward way in an environment of scale ( say more than 1000 servers ). The following problem statements are proposed:Q1: What are the concrete challenges facing administrators who are to set up and manage a cloud environment?Q2: How can todays cloud technologies be moved into more autonomic management in order to facilitate predictable and effective deployment?Q3: To what extent do current cloud technologies allow high-level management of large clusters of virtual machines?We will aim at deploying and modernising the current virtualized infrastructure in use at HiOA through this project. Throgh close investigation and experimentation with different technologies can we learn about their functional limitations. The Alto project is an umbrella project encompassing researchers from several departments at TKD. The proposed cloud would serve as a testing-ground for management approches and new tools.
Vis fullstendig beskrivelse
Metode
Det vil bli utvikling (programmering) og testing i et lab miljø. Videre vil det bli kjørt simuleringer (eksperimenter) og spesielle scenarioer, som vil danne basis for datainnsamling og analyse. Analysen vil bestå av å sammenligne resultatet av scenarioene med våre forventninger i forhold til faktorer som er viktige i en systemadministrasjonskontekst. Slike faktorer er: - Økt kontroll- Forbedret tjenestekvalitet- Raskere tilpasning av systemet til nye forhold- Implementasjon av policy- Besparelser av resursser (penger, strøm, arbeidstid osv.)
Vis fullstendig beskrivelse