Privacy-Aware Scheduling for Inter-Organizational Processes

Presentation of my research paper "Privacy-Aware Scheduling for Inter-Organizational Processes" at the ZEUS workshop 2015 in Jena, Germany.

Transcript

1. Privacy-Aware Scheduling for Inter-Organizational Processes Christoph Hochreiner Distributed Systems Group, Vienna University of Technology
 c.hochreiner@infosys.tuwien.ac.at
2. Agenda ‣ Motivation ‣ Privacy-aware scheduling ‣ Outlook 2 Motivate the need of a privacy-aware scheduling and present the approach.
3. MOTIVATION 3
4. BUSINESS PROCESS 4 Company 1 Company 2 Company 3 Service 1 Service 2 Service 4 Service 3 Service 5 privacy sensitive Service 6
5. Company 3 Company 2 Company 1 INITIAL SETUP 5 Service 1 Service 2 Service 3 Service 4 Service 5 Service 6 Company 1 Company 2 Company 3 Service 1 Service 2 Service 4 Service 3 Service 5 privacy sensitive Service 6
6. Company 3 Company 2 Community Cloud DESIRED SETUP 6 Service 1 Service 2 Service 3 Service 4 Service 5 Service 6 Company 1 Company 2 Company 3 Service 1 Service 2 Service 4 Service 3 Service 5 privacy sensitive Service 6
7. PRIVACY-AWARE SCHEDULING 7
8. FOUNDATION 8 ‣ Resource allocation and scheduling approach ‣ Appropriate SLA definition ‣ Privacy aware deployment constraints
9. RESOURCE ALLOCATION AND SCHEDULING 9 Service Instance Placement Problem [hoe2014]
 ‣ System model - Business processes - Computational resources ‣ Optimization model - Resource allocation - Scheduling of process steps - Service deployment constraints
10. SIPP ‣ Overall leasing cost ‣ Penalty cost for delayed process executions ‣ Idle resources ‣ Ordered execution of process steps 10 min X v2V cv · (v,t) + X p2P X ip 2Ip c p ip · e p ip + X v2V X kv 2Kv ( ! C f · f C kv + ! R f · f R kv ) X p2P X ip 2Ip X jip 2J⇤ ip 1 DLip ⌧t x(jip ,kv,t)
11. FOUNDATION 11 ✓ Resource allocation and scheduling approach ‣ Appropriate SLA definition ‣ Privacy aware deployment constraints
12. SLA DEFINITION 12 Requirements ‣ Simple representation of constraints ‣ Model deployment constraints Relevant approaches ‣ Blacklisting
 Optimistic approach ‣ Whitelisting
 Defensive approach
13. BLACKLISTING 13 Community Cloud Private Cloud 1 Private Cloud 2 Private Cloud 3 SLAP = 8 > < > : Community Cloud (Service 3, Service 5) Private Cloud 1 (Service 5) Private Cloud 2 (Service 3) SLAP = 8 > > > < > > > : Community Cloud (Service 3, Service 5) Private Cloud 1 (Service 5) Private Cloud 2 (Service 3) Private Cloud 3 (Service 3, Service 5)
14. WHITELISTING 14 Community Cloud Private Cloud 1 Private Cloud 2 Private Cloud 3 SLAP = 8 > < > : Community Cloud (Service 1, Service 2, Service 4) Private Cloud 1 (Service 1, Service 2, Service 4, Service 3) Private Cloud 2 (Service 1, Service 2, Service 4, Service 5) SLAP = 8 > > > < > > > : Community Cloud (Service 1, Service 2, Service 4) Private Cloud 1 (Service 1, Service 2, Service 4, Service 3) Private Cloud 2 (Service 1, Service 2, Service 4, Service 5) Private Cloud 3 ()
15. SLA DEFINITION 15 Blacklisting + Concise definition - Requires updates for new clouds Whitelisting + Requires no updates for new clouds - More detailed than blacklisting
16. FOUNDATION 16 ✓ Resource allocation and scheduling approach ✓ Appropriate SLA definition ‣ Privacy aware deployment constraints
17. PRIVACY EXTENSIONS 17 ‣ Extend the definition of available VMs V = [ loc 2 Loc V loc K = [ loc 2 Loc K loc x(j ip ,k vloc ,t) = ( 0 , if ji p / 2 SLAP loc , loc 2 Loc {0 , 1} , else ‣ Extend the definition of currently leased VMs ‣ Restrict the deployment based on the SLA
18. FOUNDATION 18 ✓ Resource allocation and scheduling approach ✓ Appropriate SLA definition ✓ Privacy aware deployment constraints Privacy-aware scheduling for Inter-Organizational Processes is possible
19. CONCLUSION 19
20. CONCLUSION 20 ‣ Evaluated different SLA approaches ‣ Formulated an approach for privacy aware 
 business process executions … but - up to now … ‣ Only a theoretical concept ‣ Needs extensive evaluation
21. OUTLOOK 21
22. OUTLOOK ‣Evaluate approach against other privacy 
 preserving approaches - Classic deployment on dedicated clouds - Other privacy ensuring methods like homomorphic encryption [gen2009] ‣Investigate data transfer aspects among clouds ‣Investigate different deployment strategies - Pricing models - Cloud/VM granularity ‣Investigate global datastores, i.e. databases 22
23. Q  &  A Christoph Hochreiner Distributed Systems Group Vienna University of Technology c.hochreiner@infosys.tuwien.ac.at