Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Cloud Economics in Training and Simulation

528 views

Published on

This slide presents a use case how to adopt IaaS cloud computing in higher education. It is shown that virtual labs can provide a more than 25 times cost advantage compared to classical dedicated on-premise in-house labs.

Published in: Technology
  • Be the first to comment

Cloud Economics in Training and Simulation

  1. 1. Nane KratzkeCLOUD ECONOMICS INTRAINING AND SIMULATION Prof. Dr. rer. nat. Nane Kratzke 1 Computer Science and Business Information Systems
  2. 2. The next 20 to 25 minutes are about ...•  What is cloud computing?•  (Economical) characteristics of cloud computing•  Postulated use cases for cloud computing•  Some data from real world•  Decision making is not always obvious => How to decide?•  Some findings Prof. Dr. rer. nat. Nane Kratzke 2 Computer Science and Business Information Systems
  3. 3. What is a cloud computing (definition)„Cloud computing is a model forenabling ubiquitous, convenient, on-demand network access to a shared poolof configurable computing resources(e.g., networks, servers, storage,applications, and services) that can berapidly provisioned and released withminimal management effort or serviceprovider interaction.“National Institute of Standards and Technology,NIST: „The NIST definition of cloud computing“;Peter Mell, Timothy Grance, 2011http://csrc.nist.gov/publications/nistpubs/800-145/SP800-145.pdf Prof. Dr. rer. nat. Nane Kratzke 3 Computer Science and Business Information Systems
  4. 4. Essential Characteristics of Clouds •  No human •  Remote access via •  Resources are interaction thin or fat client pooled to serve necessary platforms multiple consumers •  Programmable data •  No physical access •  Little control or center knowledge over exact location On-demand Network Resource self-service access pooling C enter Data •  Rapid provisioning able Pay-per-use cture amm •  gr a s tr u Pro are d Infr •  Autoscaling business model e•  Resource usage •  Ressources virtuallyware defin can be monitored, unlimited S oft controlled, and reported Rapid Measured elasticity Service Prof. Dr. rer. nat. Nane Kratzke 4 Praktische Informatik und betriebliche Informationssysteme
  5. 5. Business Characteristics Fixed costs become Pay as you go variable Cost are associative Business gains •  100 servers for one hour flexibility •  1 server for 100 hours •  no long-term financial •  (Almost) same price commitment to resources Prof. Dr. rer. nat. Nane Kratzke 5 Praktische Informatik und betriebliche Informationssysteme
  6. 6. Economical Cloud Usage Patternshave to do with peak loads „In other words, even if cloud services cost, say, twice as much, a pure cloud solution makes sense for those demand curves where the peak-to- average ratio is two-to-one or higher.“ Weinman, Mathematical Proof of the Inevitability of Cloud Computing, 2011 http://www.joeweinman.com/Resources/Joe_Weinman_Inevitability_Of_Cloud.pdf Peak loads are cloud economics best friends Prof. Dr. rer. nat. Nane Kratzke 6 Praktische Informatik und betriebliche Informationssysteme
  7. 7. Postulated use casesThese use cases (among others) are postulated to be cloud compatible: data storage, support software short-term system disaster recovery hosting websites development cycles demonstrations and business continuity overflow processing Training and media processing or large-scale simulation education and rendering scientific data processing •  Research shows that cost advantages of cloud computing are deeply use case specific •  Be aware of comparing non comparable use cases •  This contribution presents some data of educational use cases (similar usage characteristics of simulation use cases) Prof. Dr. rer. nat. Nane Kratzke 7 Computer Science and Business Information Systems
  8. 8. Analyzed use case•  Web technology lecture/practical course for computer science students (bachelor) in summer 2011 and summer/winter 2012.•  Projects: Development of web information systems (Drupal based)•  All groups were assigned cloud service accounts provided by Amazon Web Services (AWS).•  Analysis of billing as well as usage data provided by AWS. Prof. Dr. rer. nat. Nane Kratzke 8 Computer Science and Business Information Systems
  9. 9. (A) Costs per Month (aligned to Weeks) 500 Cost analysis 400Costs in USD 300 200 Total costs: 846.99 $ 100 Total students: 49 Cost per student: 17.28 $ 0 CW 13 CW 14 – CW 17 CW 18 – CW 21 CW 22 – CW 25 Calendar Weeks (CW) (B) Main Cost Drivers instancehour (62%) Main identified cost drivers: (1)  Server uptime (2/3) datatransfer (0%) adressing (3%) (2)  Data storage (1/3) datastorage (34%) Prof. Dr. rer. nat. Nane Kratzke 9 Computer Science and Business Information Systems
  10. 10. Usage Analysis (A) Maximum and Average Box Usage Training 50 Average Box Usage Maximum Box Usage in an hour 40Used Server Boxes 30 Project 24x7 Migration 20 10 0 13 14 15 16 17 18 19 20 21 22 23 24 25 Calendar Week Prof. Dr. rer. nat. Nane Kratzke 10 Computer Science and Business Information Systems (B)
  11. 11. 0 13 14 15 16 17 18 19 20 21 22 23 24 25Average to Peak Ratio per Week Calendar Week (C) Average Box to Maximum Box Ratio according to Weinman 1.0 Cloud computing is economical not reasonableAvg to Max Box Usage Ratio 0.8 Cloud computing 0.6 might be reasonable 0.4 Cloud computing is 0.2 economical reasonable 0.0 14 16 18 20 22 24 Calendar Week Prof. Dr. rer. nat. Nane Kratzke 11 Computer Science and Business Information Systems
  12. 12. Economical Decision AnalysisA four step process to decide for or against cloud based solutions (A) Determine your atp Maximum and Average Box Usage 50 ratio Average Box Usage Maximum Box Usage in an hour 40 Used Server Boxes 30 Determine your 20 dedicated costs 10 0 13 14 15 16 17 18 19 20 21 22 23 24 25 Calendar Week Determine your (B) maximal cloud costs Max instances: 49 2000 Accumulated Processing Hours per Week Processing hours: 7612 1500 Processing Hours Determine appropriate Average: 7612 / (26 * 7 * 24) = 1.74 1000 cloud ressources Overall atp ratio: 1.74 / 49 = 0.035 500 Prof. Dr. rer. nat. Nane Kratzke 12 0 Computer Science and Business Information Systems 13 14 15 16 17 18 19 20 21 22 23 24 25
  13. 13. Economical Decision AnalysisA four step process to decide for or against cloud based solutions Determine your atp „In other words, even if cloud services cost, ratio say, twice as much, a pure cloud solution makes sense for those demand curves where the peak-to-average ratio is two-to-one or higher.“ Determine your dedicated costs Weinman, Mathematical Proof of the Inevitability of Cloud Computing, 2011 Determine your Example Server: 500 US Dollar maximal cloud costs Amortization: 3 years 500$ Determine appropriate d3years (500$) = = 0.019 $ h cloud ressources 3 • 365 • 24h Prof. Dr. rer. nat. Nane Kratzke 13 Computer Science and Business Information Systems
  14. 14. Economical Decision AnalysisA four step process to decide for or against cloud based solutions According to Weinman the peak-to-average Determine your atp ratio ratio should be greater than the ratio between the variable costs c and your (assumed) dedicated costs d: Determine your dedicated costs Determine your maximal cloud costs Determine appropriate 0.019 $ h $ cloud ressources c Max = = 0.54 0.035 h Prof. Dr. rer. nat. Nane Kratzke 14 Computer Science and Business Information Systems
  15. 15. Economical Decision AnalysisA four step process to decide for or against cloud based solutions Determine your atp 0.019 $ h $ ratio c Max = ≈ 0.54 0.035 h Pricings for EU region, 19th March, 2012 Determine your € Example: Amazon Web Services EC2- dedicated costs Determine your maximal cloud costs Determine appropriate cloud ressources Prof. Dr. rer. nat. Nane Kratzke 15 Computer Science and Business Information Systems
  16. 16. Economical Decision AnalysisA four step process to decide for or against cloud based virtual labs The measured ATP ratio of 0.035 means in fact a 1/0.035 == 28.57 times cost advantage. This means for the presented use case: A cloud based solution provides a more than 25 times cost advantage. Compared to necessary investment efforts for a classical dedicated system implementation. Prof. Dr. rer. nat. Nane Kratzke 16 Computer Science and Business Information Systems
  17. 17. Why this big cost advantage? (A) How to dimensionize the data center? Maximum and Average Box Usage peak load 50 Average Box Usage And the delta? Maximum Box Usage in an hour 40 Used Server Boxes 30 Measures the overdimension of a data center 20 average 10 load 0 13 14 15 16 17 18 19 20 21 22 23 24 25 Calendar WeekWhat is the need? Prof. Dr. rer. nat. Nane Kratzke 17 Computer Science and Business Information Systems (B)
  18. 18. In other words ... (A) Maximum and Average Box Usage You have to finance a really big house ... 50 Average Box Usage Maximum Box Usage in an hour 40 ... knowingUsed Server Boxes that you will inhabit 30 only some rooms of it. 20 10 0 13 14 15 16 17 18 19 20 21 22 23 24 25 Calendar Week Prof. Dr. rer. nat. Nane Kratzke 18 Computer Science and Business Information Systems (B)
  19. 19. Findings •  Cloud computing loves peak load scenarios (be happy) •  25 times cost advantage (analyzed use case) •  Cloud generated costs are use case specific (be carefull) •  Decision making must not be obvious •  Four step decision making model (to determine your ATP ratio) •  Main cost drivers are (try to minimize) •  Server uptime •  Data storage (server volumes) •  Data transfer (in communication intensive use cases) •  Uneconomical use cases (try to avoid) •  24x7 and •  constant loads •  So if you have to deal with peak load scenerios it is likely that cloud based solutions might be an economical option ... Prof. Dr. rer. nat. Nane Kratzke 19 Computer Science and Business Information Systems
  20. 20. Thank you for listening Find this presentation here: http://www.slideshare.net/i21aneka/itis-ws-2013 Slideshare: i21aneka XING: Nane_Kratzke LinkedIn: nanekratzkeProf. Dr. Nane KratzkeComputer Science and WEB:Business Information Systems http://praktische-informatik.fh-luebeck.deLübeck University of Applied SciencesMönkhofer Weg 239 Mail: Twitter:23562 Lübeck nane.kratzke@fh-luebeck.de @nanekratzkeGermany Prof. Dr. rer. nat. Nane Kratzke 20 Computer Science and Business Information Systems
  21. 21. Qualitative IT-Management Impact of Clouds Governance Enterprise system design Operation (COBIT) (TOGAF) (ITIL) 12 x Positive 3 x Positive 6 x Positive 8 x Negative 0 x Negative 3 x Negative Prof. Dr. rer. nat. Nane Kratzke 21 Computer Science and Business Information Systems
  22. 22. Advantages and short comings of cloud computing Advantages Short comings l cture and low leve Physical infrastru mer perspective) o service free (cust mpliancy More complex co t ted functional managemen Pro vision of automa services t cu rity managemen More complex se (ex post) Cost transparency d rvice, process an More complex se anagement ntinuousity and configuration m Inhe rent scalability, co availability Prof. Dr. rer. nat. Nane Kratzke 22 Computer Science and Business Information Systems
  23. 23. So – everthing is beautifull?No substantial show stoppers?•  Higher order showstoppers for cloud approaches Hard to handle •  Security and Compliance Management •  Incompatible SLAs •  Especially national laws, privacy, data ownership, confidentiality, data location, forensic evidence, auditing, etc.•  Decision making showstoppers for cloud approaches Could be solved •  Ex post but no ex ante cost transparency •  Relevant costs of cloud approaches must be known before a system enters operation •  Otherwise IT investment decisions pro or contra cloud based approaches can not been made Prof. Dr. rer. nat. Nane Kratzke 23 Computer Science and Business Information Systems
  24. 24. Typical Cost Structure Infrastructure ... Platform ... Software ... ... as a Service Service Level Cost category •  IaaS + Scalability •  datatransfer •  PaaS •  dataprocessing •  SaaS •  datastorage •  network •  monitoring •  per request •  per user/account Prof. Dr. rer. nat. Nane Kratzke 24 Computer Science and Business Information Systems
  25. 25. Assignment of cost categories to Cloud Service Levels Per Data Data Data Net- Moni- Request/ storage processing transfer work toring UserScalability X X X X IaaS X X X X X (per micro request) X PaaS X X X (per request) X SaaS X X (per user) Prof. Dr. rer. nat. Nane Kratzke 25 Computer Science and Business Information Systems

×