(MC²)²: A Generic Decision-Making Framework and its Application to Cloud Computing
1. (MC²)²: A GenericDecision-Making Framework anditsApplicationtoCloud Computing Michael Menzel1, Marten Schönherr2, Jens Nimis1, Stefan Tai31FZI Karslruhe, 2T-Labs Berlin, 3KIT Karslruhe
2. “Which IT infrastructure?“ iscomplex Existingapproachesonlyconsidercostaspects! Operationsresearch: Ngwenyama/Bryson[1] proposetransactioncosts Cloudspecific Armbrust2:cpuhourcostsofcloud vs. Datacenter Cloudspecific Walker3: netpresentvalueofleasingorbuying CPU Considering also non-costfactorsrequiresextensionofexistingapproaches => multi-criteriadecision-making (MCDM): Multi-criteriadecision-makingaloneisnot sufficient !=> supportandcustomizationof multi-criteriadecision-making in a framework Motivation of (MC²)² [1] Ngwenyama, O. & Bryson, N. (1999), 'Making the information systems outsourcing decision: A transaction cost approach to analyzing outsourcing decision problems', European Journal of Operational Research 115(2), 351-367. [2] Armbrust, M.; Fox, A.; Griffith, R.; Joseph, A.; Katz, R.; Konwinski, A.; Lee, G.; Patterson, D.; Rabkin, A.; Stoica, I. & others (2009), 'Abovetheclouds: A berkeleyviewofcloudcomputing', EECS Department, University ofCalifornia, Berkeley, Tech. Rep. UCB/EECS-2009-28. [3] Walker, E. (2009), 'The Real Cost of a CPU Hour', Computer 42, 35-41.
9. Summary: New CRM softwareforExamChoccompany Mustbecostefficient Mustbeverysecure Loadprofileknown (roughly) Support green IT Example – scenario The scenario of this example shall address the widely articulated IT infrastructure decision-making problem. Consider a chocolate producing company called ―ExamChoc‖, which needs to make an IT infrastructure decision on the deployment of a new Customer Relationship Management (CRM) software system. After a long analysis by the company’s chair it is decided that the solution has to be cost efficient and very secure at the same time. The chair based the decision on the current economic situation the organization has to cope with and on the company’s regulations that require a high security level for customer data. Due to the nature of ExamChoc’s business the usage statistics of the CRM software are forecasted to be very seasonally varying. Also, the software will not be in use between 6 pm and 8 am as very strict policies for working hours are set upon the employees. Additionally, the company cares about its corporate responsibility strongly and, thus, is supporting green IT approaches whenever possible.
11. C1. Security & Reliability C1.1. Data safety level of the infrastructure C1.2. Reliability of the infrastructure C1.3. Data security level in the infrastructure C1.4. Security level of data transmissions from and to the infrastructure C2. Benefits C2.1. Number of resources available C2.2. Green IT: power consumption during usage C3. CloudTCOcriteria C3.1. Uptime costs for dynamic usage C3.2. Costsforsecurity C3.3. Initial upfrontcosts C3.4. Administration costs C3.5. Costs for internet connection C3.6. Research costs for hardware Example – criteria
12. R1. Trustworthiness of the provider R2. Green IT: minimum power consumption per hour per server Example - requirements
13. Wechoose: ANP – Analytic Network Process Supports complexcriteriadependencies Considers qualitative criteria Results on absolute scale (0-1) in othercasesothermethodsmight fit better MAUT Promethée … Example – choose multi-criteriadecision-makingmethod
20. (MC²)² canbeimprovedandextended in manyaspects: More IT supportwithintheframework List ofsuitableinformationsystemsthatcansupportthedefinitionof alternatives, criteria, andrequirements Prototype of a systemthatoffers (MC²)² decisionmakingandintegrationwithinformationresources Inclusionofsensitivityanalysisandgroupdecisionaspects Support withthechoiceof MCDM methods … Future work
26. To find alternatives, criteria, requirements, and MCDM method Expert searchandsurveycreationsystems Internal knowledgeordatabaseswith Experiencesfromearlierevaluations Guidelinedocumentsofdepartments Collaborationsystems Data miningtools Literature … Toexternalizeyourexperiencesmadeduringtheuseoftheframework Useofinformationresources…
27. Qualitative vs. Quantitative (non-measurable vs. measurable) Scalesofmeasurement Rules toconsider Completeness and Exhaustiveness Mutually exclusive items only Restrict to criteria of highest degree of importance Examples: Criteria
29. Manymethodsavailable* Outranking: ELECTRE, PROMETHEE TOPSIS, AHP/ANP Bayesian Analysis MAUT, SMART … Chooseoneandconsider Not everymethodresults in a value Not everymethodfitsyourneedsto model criteriadependencies Not everymethodsupports qualitative (non-measurable) criteria well Multi-criteriadecision-makingmethod * Yoon and Hwang (1995), ‘Multi Attribute Decision Making: An Introduction‘, Sage University Paper series on Quantitative Applications in theSocialSciences. Keun Tae Cho (2003), ‘Multicriteria Decision Methods: An Attempt To Evaluate and Unify’, Mathematicaland Computer Modelling. Fülöp (2004), ‘Introduction to Decision Making Methods’, Paper developed for the BDEI-3 Workshop, The Evergreen State College.