The OCCIware project aims to lower cloud computing adoption costs and break barriers between implementations through formalizing and extending the Open Cloud Computing Interface (OCCI) standard. The project developed an Eclipse-based, model-driven platform for managing cloud resources across multiple domains. It addressed challenges including the need for open standards, hybrid cloud elasticity, data-centric clouds, and cloud management. The 72 person-month, 5.6M Euro project involved 3 academics and 5 companies and resulted in a formal OCCI model, tools, and use cases spanning infrastructure, big data, deployment, and linked data.

1. OCCIware Open Standards
Challenges
Iyad Alshabani
ActiveEon
Contribution on the
“Consultation on Cloud Computing Research
Innovation Challenges”

2. OCCIware Factsheet
• 72 man month, 5,6m€ budget
• 3 academics, 5 companies, 2 associations
• To lower Cloud Computing adoption costs and
break up barriers between its various
implementations, layers, domains
- Especially Data Center, deployment, Big Data, Linked
Data
• By bringing to the Open Cloud Computing Interface
(OCCI) the power of formal languages and model
driven engineering (MDE)

3. Cloud Computing – the problem
Partitioning! Lock in!
No one-fits-all
solution!
Lack of standards!
Layers
Domains
Too many
standards!
Lack of ecosystem!

4. OCCIware output
• A formal, model-driven platform to manage any Cloud
resource
- Formal model of OCCI (Open Cloud Computing Interface)
- Eclipse tooling – OCCIware Studio
- Models@runtime
- Deploy@OCCIware
• 4 Use cases
- Data Center as a Service / IaaS
- Big Data / HPC
- Linked Data
- Deployment interoperability
• Open Source (OW2, Eclipse) and standardization (OGF)
Brussels, 07/11/2016 6Iyad Alshabani

5. Partners and Contributions
• Inria – formal model, Scientific Coordinator
• Obeo – modelers, tooling
• IMT / TSP – simulation, OCCI experience
• ActiveEon - HPC / Big Data, deployment (workflow)
• Scalair : Data Center, administration
• Linagora : deployment, monitoring
• UJF : deployment (dependancies, optimization)
• Open Wide/Smile : Linked Data core, Project Coordinator
• OW2 : Open Source community
Brussels, 07/11/2016 7Iyad Alshabani

6. Challenges
• Standards and Openness
• Elasticity
• Hybrid cloud
• Extended Cloud (P2P, Fog, Edge)
• Data-Centric Clouds
• Cloud Management
Brussels, 07/11/2016 8Iyad Alshabani

7. Standards and Openness
• Theory of Cloud Computing
• Various standards
• OVF, DMTF, CIMI, OASIS CAMP, OGF OCCI, OASIS TOSCA
• Each one covers only one part of cloud interoperability
• Not supported by all cloud offers
• Cloud computing is a collection of informal and
ambiguous definitions
• Lack in theory
• To provide formal definition
• Prove the foundation, semantic interoperability
• Capture and specify characteristics
• For IaaS, PaaS, SaaS .. And Everything aaS (XaaS)
Brussels, 07/11/2016 Iyad Alshabani 9

8. Standard and Openness
• Further cloud standards
• Open APIs and open implementations
• To avoid vendor lock-in
• More interoperability
• Extended cloud to Fog/Edge
• EU Open Source can compete US software and service
providers
Brussels, 07/11/2016 Iyad Alshabani 10

9. Elasticity
• Thresholds definition and prediction estimation
• Hybrid elasticity (VMs and Containers)
• optimal tradeoff between user requirement and
providers interests
• Cost-effective and flexible Cloud resource allocation
• Including the Edge with predictive workload
management
• Elasticity “à la carte”
• One-size-fits-all solution
• Elasticiy=Scaling+Automtion+Optimization
Brussels, 07/11/2016 Iyad Alshabani 11

10. Hybrid Cloud
• Private, Public, but also Edge and P2P
• ProActivity in resource management
• Predictive and Adaptive monitoring
• Optimal Things provisioning in Cloud/Fog
• Security, scalability, elasticity and continuous
deployment
• Cloud migration (especially at PaaS level)
• Devops, continuous delivery
Brussels, 07/11/2016 Iyad Alshabani 12

11. Data Centric Cloud
• Real-time big data processing
• Cloud/Edge Computing
• Adaptive Cloud resource management
• Resource allocation for big data applications
• Big Data workload management
• Scalable, federated management of Heterogenous
data and models
• Open data access
• Data Analytic deployment decision making
• Unified view of data processing
Brussels, 07/11/2016 Iyad Alshabani 13

12. Cloud Management
• Self management
• Controllable SLA
• Monitoring
• All levels and traversal (IaaS, PaaS, SaaS, XaaS)
• Distributed run-time support
• Virtual machine
• collapsing the different virtualization layers
• Devops tools
Brussels, 07/11/2016 Iyad Alshabani 14

13. Thank you
OCCIware project home page
http://occiware.org/
Brussels, 07/11/2016 Iyad Alshabani 15