1. University of Stuttgart
Decision support for partially moving applications to cloud environments The example of Business Intelligence
Adrián Juan-Verdejo
Dr. Henning Baars
University of Stuttgart
and
CAS Software AG
21st of April 2013

2. Motivation Decision-support system Partial migration
Legacy applications can leverage cloud
computing: scalability, availability, cost...
Adapting an application is a complicated decision-making process
Many factors
Interdependent factors
Security and privacy
QoS requirements
Decision-support
system
--Motivation
Legacy applications can benefit from the use of Cloud computing technologies: scalability, availability, cost...
However, complicated decision-making process: many interdependent factors
respect to SLA, data sensitivity, privacy concerns, national or enterprise-specific regulations, increased transactional delays, spread and variability of users, increased wide-area communications costs, costs related to computation and storage, law, etc.
Partial migration
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

3. Partial Migration Migration Process
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

4. Goal and Challenges Goal Challenges
Assist the partial migration of legacy applications to cloud computing environments
Methodology and decision support system
Provide the needed functionality
Respect different set of requirements
Privacy and security-related requirements
QoS requirements
Component’s interdependencies
Select the right cloud provider
Still be economically beneficial
Goal
Challenges
respect to SLA, data sensitivity, privacy concerns, national or enterprise-specific regulations, increased transactional delays, spread and variability of users, increased wide-area communications costs, costs related to computation and storage, law, etc
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

5. Topics Motivation & Challenges Research Gap
Approach: Moving applications to cloud env.
Moving BI to cloud environments
Conclusion & Acknowledgements
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

6. Research Gap (1/2)
Research work
Focus
Granularity
Application's Setting
Moving-to-the-cloud problem[4]
Moving apps. To cloud envs.
Components
Pre‑existing applications
Cloudward Bound [6]
MCDM migration
Enterprise App
Volley [11]
Data partitioning
MapReduce Jobs
MapReduce
Manticore [12]
Code partitioning
Code entities
Software services
HybrEx [13]
Data and system partitioning focusing on privacy
Distributed Applications
COPE [14]
Automated orchestration using declarative languages
Virtual Machines
CloudGenious [15]
Web servers MCDM migration
Web Apps
(MC2)2 [16]
Conceptual framework
Conductor [17]
Orchestration deployment
Multi-criteria decision analysis for the migration of components within a legacy enterprise application
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

7. Research Gap (2/2)
Holistic multiple-criteria decision-making approach to find the best suitable deployment
Heterogeneous and interdependent user’s requirements: QoS, privacy, security, business, economics
Legacy application’s architecture
Selection of the right cloud provider
respect to SLA, data sensitivity, privacy concerns, national or enterprise-specific regulations, increased transactional delays, spread and variability of users, increased wide-area communications costs, costs related to computation and storage, law, etc
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

8. Topics Motivation & Challenges Research Gap
Approach: Moving applications to cloud env.
Moving BI to cloud environments
Conclusion & Acknowledgements
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

9. Approach: Moving applications to cloud env.
Description of pre-existing system
Data sensitivity
8:42
Architecture
Dependencies
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

10. Approach: Moving applications to cloud env.
Division and scattering of components
Max Benefits (M) – InternetCosts(M)
Subject to Privacy, security, gobernance policies
QoS requirements
Pre‑existing interdependencies
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

11. Approach: Moving applications to cloud env.
Multi-criteria decision-making
Analytical Hierarchy Process (AHP) [38] … …
---problem decomposition:hierarchy structure that specifies the interrelation among three kinds of elements
---judgment of priorities:
pairwise comparisons of attributes are done to specify their relative priorities.
pairwise comparison of Cloud services and deployment is done based on attributes to compute their local ranks,
---aggregation of these priorities
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

12. Topics Motivation & Challenges Research Gap
Approach: Moving applications to cloud env.
Moving BI to cloud environments
Conclusion & Acknowledgements 15
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

13. What is BI?
Integrated and multi-layered IT-based management and decision support
Cloud-BI
Heterogeneous components
Different requirements
Interdependent
Data
Logic: reporting, data mining, and OLAP tools
Access: usually a portal
theories, methodologies, processes, architectures, and technologies that transform raw data into meaningful and useful information for business purposes
Baars, H. and H.G. Kemper, Business Intelligence in the Cloud?, in 14th Pacific Asia Conference on Information Systems (PACIS), 2010, Taipeh, Taiwan.
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

14. Scenario 1: data analysis functionality
Inclusion of specialized data analysis functionality on CC
Interdependencies with data
Data sensitivity
Volume of data to be moved
Cost
Inclusion of specialized data analysis functionality
Interdependencies with data
data sensitivity: market analysis based on web site data
Volume of data to be moved
Cost (moving, running there)
E.g. Market analysis
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

15. Scenario 2: Move OLAP frontend
Move a reporting or Online Analytical Processing frontend
Security & privacy
Data updates
Consistency
Cost
Performance
2-----OLAP: tool-based approach. Move reporting or/and OLAP
More accesibility (mobile devices for example)
implications for reaction times to the user's input and data traffic. CONSISTENCY problems
LESS Far from rest of BI (delay MOVE data mart DWH
The movement of the data mart might alleviate the reaction time issues but has further implications regarding security (not only a few reports are exposed to the web but the whole data repository)
Security & privacy (when we move data mart=, consistency
Updates of data marts: bottlenecks in cost and performance leading to move DWH to the Cloud
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

16. Scenario 3: Move Operational BI
Sometimes called real-time BI
Decisions based on real-time data e.g. call centre
Move an Operational BI solution
BI triggers events in other systems (active BI)
Data updates in both directions
Where to place functionality to trigger events
--Move an Operational BI solution
CRITICAL Active BI: it triggers events in oher systems : INTERDEPENDENCIES
Data updates in both directions
front-line workers, such as call centres operators, who need timely data to do their jobs.
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

17. Scenario 4: Move Selected ETL procedures
Move selected Extract-Transform-Load procedures
Data sources in cloud environments
Diminish traffic
Routines into ETL before feeding data to DWH
Pre-processing unstructured data
Discovering non-evident duplicate entries
Move selected ETL procedures
Extracting data from outside sources
Transforming it to fit operational needs, which can include quality levels
Loading it into the end target (database, more specifically, operational data store, data mart or data warehouse)
particularly for data sources in cloud env.
-----data sources in the master data before is fed into the data warehouse
In this case, the routines have to be embedded into the higher-order ETL process. A lot of data, ETL link several core components => affect the overall BI on various levels.
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

18. Topics Motivation & Challenges Research Gap
Approach: Moving applications to cloud env.
Moving BI to cloud environments
Conclusion & Acknowledgements
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

19. Conclusion and Acknowledgements
Generic cloud migration framework
interdependent user’s requirements: QoS, privacy, security, business, economics
application’s architecture
selection of the right cloud provider
Apply it to the real-case scenario of BI
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

20. Questions?
Adrián Juan-Verdejo University of Stuttgart and CAS Software AG
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

21. Future steps Continue collaboration @Stuttgart
Decision model refinement
Further characterize criteria
Apply DM to the identified BI scenarios
Experiment with EMF tools for legacy application description
Incorporate cloud alternatives specifics SMICloud
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

22. Papers since mid-term Papers HotTopics with Vahid and Bholanath
Juan-Verdejo A., Baars, H. Decision support for partially moving applications to the Cloud – the example of Business Intelligence. In Proceedings of the International Workshop on Hot Topics in Cloud Services (HotTopiCS 2013) within the 4th ACM/SPEC International Conference on Performance Engineering (ICPE 2013), April 18-24, Prague, 2013.
To be submitted
Refinement of topic
Paper with Vahid and Bholanath
Adrián Juan-Verdejo, Bholanathsingh Surajbali, Seyed Vahid Mohammadi, Henning Baars, and Hans-Georg Kemper. Moving Business Intelligence to cloud environments: A security-enhanced Framework
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

23. Cloud migration process
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

24. Approach: Moving applications to cloud env.
Division and scattering of components
Max Benefits (M) – InternetCosts(M)
Subject to Privacy, security, gobernance policies
QoS requirements
Pre‑existing interdependencies
Decision support for partially moving applications to the Cloud Adrián Juan-Verdejo

25. BI system running within a Cloud Provider
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

26. CMI Qualitative Parameters (NNci)
Sensitivity of the data to be migrated
Regulated data, most sensitive. Credit card numbers, bank accounts, driver's license, health information...
Confidential data, high sensitivity. Personnel information, financial information, contracts
Public data, low sensitivity. Maps, publicly available product info
Is there any framework for this?
Level 1—Low Sensitivity
Information at this level requires a minimal amount of protection. This level includes information that is considered to be in the public domain, such as employee locator files. At this level, any disclosures could be reasonably expected not to have an adverse effect. But remember that all information is important, otherwise it would not be collected.
Unintentional alteration or destruction is the primary concern for low sensitivity information.
Level 2—Moderately Sensitive
Level 2 or Moderate Sensitivity includes data that are important to NIH, and therefore must be protected against acts that are considered to be malicious and destructive. However, disclosure problems are usually not significant since this type of data is often collected for analytical reasons.
This level includes information that pertains to workload, staffing, correspondence, memoranda, and other document files whose release or distribution outside the federal government and/or within NIH needs to be controlled. Access to Level 2 data needs to be restricted only to a limited degree. The data must be protected from unauthorized alteration or modification due to its value to the organization; however, it may be disclosed in some format eventually.
Moderately sensitive data can include information that must be protected to meet Privacy Act requirements. At this level, unauthorized disclosures could cause embarrassment to an individual.
Level 3—High Sensitivity
Everyone at NIH should be most aware of the protection requirements for Level 3 or High Sensitivity information. This level covers the most sensitive information at NIH and requires the greatest security safeguards at the user level.
This data could include computerized correspondence and document files that are regarded as highly sensitive and/or critical to an organization, and therefore must be protected from unauthorized alteration, modification, and/or premature disclosure; proprietary information that has inherent informational value, such as drug formulas, trade secrets, and early research findings; financial data that is used to authorize or make payments to individuals or organizations; clinical trial data; grant application review data; automated systems or records subject to the Privacy Act for which unauthorized disclosure would constitute a clearly unwarranted invasion of personal privacy.
Highly sensitive data must be protected from unauthorized disclosure.
Level 4—High Sensitivity and National Security
This level of data does not apply to NIH.

27. Cloud Migration Index Hard conditions Relative Condition Conclusion
Policiy constraints not satisfied
Not migrated
Constraints on delay/latency increase exceeded
Condition
Conclusion
Criteria
+ or – (or depends)
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

28. Cloud Migration Index
Assesses the suitability of a component to be migrated to a cloud environment:
Calculate quantitative values
Pair-wise comparisons (or other methods) to calculate the qualitative values
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

29. Framework Prototype (0/3)
Cloud provider and VM image selected
Specify its characteristics as they affect the criteria to move components or data
SMICloud: Quality model based on SMI framework
STRATOS:
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

30. Framework Prototype (1/3)
Define legacy application, application components’ requirements, and components’ hard constraints: policies and performance

31. Framework Prototype (2/3)
EMF-based Frameworks
CAFE
VbMF
Present new deployment to the user by using the same graphical tools

32. Framework Prototype (3/3)
Estimate system’s behaviour after migration
Tune ranking
Decision-making process related to migration options

33. Bibliography
Armbrust, M., et al., A view of cloud computing, Communications of the ACM, 2010, Vol. 53 (4), p
Khajeh-Hosseini, A., et al., Decision support tools for cloud migration in the enterprise, in IEEE International Conference on Cloud Computing (CLOUD), 2011: IEEE.
Rygielski, P. and S. Kounev, Network Virtualization for QoS-Aware Resource Management in Cloud Data Centers: A Survey, Praxis der Informationsverarbeitung und Kommunikation, 2013, Vol. 36 (1 ), p. 55–64.
Leymann, F., et al., Moving applications to the cloud: an approach based on application model enrichment, International Journal of Cooperative Information Systems, 2011, Vol. 20 (03), p
Rizou, S.a.P., A., Towards value-based resource provisioning in the Cloud, in IEEE 4th International Conference on Cloud Computing Technology and Science, 2012.
Hajjat, M., et al., Cloudward bound: planning for beneficial migration of enterprise applications to the cloud, 2010: ACM.
Moss, S.T. and S. Atre, Business Intelligence Roadmap – The Complete Project Lifecycle for Decision Support Applications, 2003, Boston: Addison-Wesley
Turban, E., R. Sharda, and D. Delen, Decision Support and Business Intelligence Systems, 9th edition ed, 2010, Bosten u.a.: Pearson.
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

34. Bibliography
Zimmer, M., H. Baars, and H. Kemper, The Impact of Agility Requirements on Business Intelligence Architectures, in System Science (HICSS), th Hawaii International Conference on, 2012: IEEE.
Thompson, W.J.J. and J.S. van der Walt, Business intelligence in the cloud, SA Journal of Information Management, 2010, Vol. 12 (1), p. 5 pages.
Agarwal, S., et al., Volley: Automated data placement for geo-distributed cloud services, in Proceedings of the 7th USENIX conference on Networked systems design and implementation, : USENIX Association.
Kaviani, N., E. Wohlstadter, and R. Lea, MANTICORE: a Framework for Partitioning Software Services for Hybrid Cloud, in 4th ieee international conference on cloud computing technology and science2012: Taipei, Taiwan.
Ko, S.Y., K. Jeon, and R. Morales, The hybrex model for confidentiality and privacy in cloud computing, in Proceedings of the 2011 conference on Hot topics in Cloud Computing. USENIX Association, Portland, OR, 2011.
Liu, C., B.T. Loo, and Y. Mao, Declarative automated cloud resource orchestration, in Proceedings of the 2nd ACM Symposium on Cloud Computing, 2011: ACM.
Menzel, M. and R. Ranjan, CloudGenius: Automated Decision Support for Migrating Multi- Component Enterprise Applications to Clouds, in International World Wide Web Conference Committee (IW3C2), 2011.
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

35. Bibliography
Menzel, M., et al., (MC2) 2: A Generic Decision-Making Framework and its Application to Cloud Computing, Software: Practice and Experience, 2011, Vol.
Wieder, A., et al., Orchestrating the Deployment of Computations in the Cloud with Conductor, in Proceedings of the 9th USENIX conference on Networked Systems Design and Implementation, 2012: USENIX Association.
Chong, S., et al., Building secure web applications with automatic partitioning, Communications of the ACM, 2009, Vol. 52 (2), p
Hunt, G.C. and M.L. Scott, The Coign automatic distributed partitioning system, Operating systems review, 1998, Vol. 33, p
Apache-Foundation, The Apache Hadoop project., 2013, Last Update, Date, Available from:
Baars, H. and H.G. Kemper, Management Support with Structured and Unstructured Data – An Integrated Business Intelligence Framework, Information Systems Management, 2008, Vol. 25 (2), p
Shollo, A. and K. Kautz, Towards an Understanding of Business Intelligence (Paper 86), in Australasian Conference on Information Systems 2010 (ACIS 2010), 2010, Brisbane (Australia).
Negash, S., Business Intelligence, Communications of AIS, 2004, Vol. 13, p
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

36. Bibliography
Golfarelli, M., S. Rizzi, and I. Cella, Beyond data warehousing: what's next in business intelligence?, in Proceedings of the 7th ACM international workshop on Data warehousing and OLAP, 2004: ACM.
Marjanovic, O., The Next Stage of Operational Business Intelligence: Creating New Challenges for Business Process Management, in 40th Annual Hawaii International Conference on System Sciences (HICSS), 2007: IEEE.
Manyika, J., et al., Big data: The next frontier for innovation, competition, and productivity, 2011.
Jacobs, A., The pathologies of big data, Communications of the ACM, 2009, Vol. 52 (8), p
Gutierrez, N., Business Intelligence (BI) Governance, Infosys White Paper, 2006.
Baars, H. and H.G. Kemper, Business Intelligence in the Cloud?, in 14th Pacific Asia Conference on Information Systems (PACIS), 2010, Taipeh, Taiwan.
Baars, H. and H.G. Kemper, Ubiquitous Computing–an Application Domain for Business Intelligence in the Cloud?, 2011, Vol.
Thomson, W.J.J. and J.S. van der Walt, Business intelligence in the cloud South African Journal of Information Management, 2010, Vol. 12 (1).
Qie, L. and H. Baars, Die Cloud als neuer Ansatz zur Erhöhung der BI-Agilität?, BI Spektrum, 2012, Vol. 7 (2), p
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

37. Bibliography
Gartner, Gartner Says Nearly One Third of Organizations Use or Plan to Use Cloud Offerings to Augment Business Intelligence Capabilities, 2012, Last Update, Date [cited th of January 2012], Available from:
Zeleny, M., Multiple criteria decision making, Vol. 25, 1982: McGraw-Hill New York.
Saaty, T.L., Theory and applications of analytic network process, Vol. 4922, 2005: RWS publications Pittsburgh.
Tran, V.X., H. Tsuji, and R. Masuda, A new QoS ontology and its QoS-based ranking algorithm for Web services, Simulation Modelling Practice and Theory, 2009, Vol. 17 (8), p
Garg, S.K., S. Versteeg, and R. Buyya, Smicloud: A framework for comparing and ranking cloud services, in Utility and Cloud Computing (UCC), 2011 Fourth IEEE International Conference on, 2011: IEEE.
Bain, Henry, Nigel Howard, and Thomas L. Saaty. Using the analysis of options technique to analyze a community conflict. Journal of Conflict Resolution (1971):
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo

38. todo Slide 1 Slide 2 Check spelling Check footnotes Change name
Decision support for partially moving applications to the Cloud: The example of BI Adrián Juan-Verdejo