Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Cloud Computing: From Revolution to Evolution
1. Cloud Computing:
From «Revolution» to «Evolution»
1 1,2 1
Sébastien Mosser, Eirik Brandtzæg, Parastoo Mohagheghi,
(1) SINTEF IKT,
(2) University of Oslo
BENEVOL’11, Brussels, Belgium
December 8th 2011
2. Cloud-Computing: From Ads ...
«Much like plugging in a microwave in order
to power it doesn’t require any knowledge of
electricity, one should be able to plug in an
application to the cloud in order to receive
the power it needs to run, just like a utility.»
http://jineshvaria.s3.amazonaws.com/public/cloudbestpractices-jvaria.pdf 2
3. ... To Reality!
«However, we are not there yet.»
3
http://jineshvaria.s3.amazonaws.com/public/cloudbestpractices-jvaria.pdf
4. Outline
• Background
• «Clouds, Clouds, Clouds ...»
• Migrating to the Clouds
• The REMICS Project
• Towards PIM4Cloud
• A language to support evolution to the cloud
• Perspectives & Conclusions
4
8. SaaS
PaaS
Resources? «*aaS» approaches
IaaS
• Infrastructure as a Service (IaaS)
• Virtual images (e.g., Linux, Windows) running in the clouds
• Platform as a Service (PaaS)
• Technological stack (e.g., J2EE, Azure) deployed in the clouds
• Software as a Service (SaaS)
• Business application (e.g., word processor) hosted in the clouds
Cloud Providers handle the «technical stuff» 8
10. Example: Launching a new product
reduced initial investment
load o pe
H
zz
Bu
private
beta scaling
present NOW future time 10
11. Example: Launching a new product
reduced initial investment
load o pe
H
stabilization
zz
Bu
private
beta scaling
present NOW future time 10
12. Example: Launching a new product
reduced initial investment
load o pe
H
stabilization
zz
could be worse ...
Bu
private
beta scaling crash
present NOW future time 10
13. Migrating to the Clouds
Adaptive Maintenance in the REMICS project
inside
14. Facts
• Funding: STREP FP7, Objective 1.2 call 5
• Internet of Services & Software Virtualization
• Started in September 2010, 3 years timeline (4.5 M€)
• Initial Consortium:
• Norway: SINTEF (leader), DI-Systemer
• France: Softeam, Netfective
• Germany: Franhofer FOKUS
• Spain: ESI Tecnalia, DOME
12
15.
From «Legacy» to «the Cloud»
Legacy
Cloud
System
13
18. CloudS StandardS ...
• Focus on Infrastructure modeling
• Useful for Cloud Provider
or
• Focus on Business viewpoint
• Useful for Business Analyst
Nothing in
between ?? 16
http://cloud-standards.org
20. PIM4Cloud
Needs: Distributed Systems
• Introducing cloud-specific concepts in deployment models
• Platform independence (avoids vendor lock-in)
• Transparent projection to multiple clouds
• Automated deployment
18
21. PIM4Cloud
Use-case driven development
• PIM4Cloud design is driven by our use cases
• Empirical approach: Model relevant concepts used in the case studies
• Initial cases:
• ERP Accounting (DI Systems, NO), Tourism (DOME, ES)
• Extended consortium:
• Bank (WUT, PO), Scientific Apps (UT, ET), Transport (IIICT, BG)
19
22. PIM4Cloud
Needs: Software Evolution
• Migrating software to the cloud
• Migration patterns? Methodologies? Impact of the cloud?
• Supporting the evolution of cloud-based applications
• Migrating to the cloud is «just» the start of something else ...
Cloud
20
23. PIM4Cloud
Evolving to the cloud ...
• Migrating to the cloud induces a paradigm shift
• More than «distributed» ... «cloudified».
• How cloud-specificities interacts with the evolution process?
• Supporting evolution of cloud-based applications:
• Impact of a given evolution on cloud-awareness
• How the evolution process interacts with cloud-specificities?
How to propagate the evolution into multiple instances,
deployed in heterogeneous and multiple clouds? 21
25. PIM4Cloud
PIM4Cloud: Prototype status
• The language is still at an early stage of development
• Will be available on GitHub in a couple of weeks
• Technological choice:
• Internal DSL in Scala
• Multi-cloud wrapper
• UML profile for the link with the UML
• On-going case studies:
• Bank management, Sensor networks
23
26. PIM4Cloud
Perspectives & Ongoing Work
• Short-term perspective:
• From «proof of concept» to «prototype»
• Mid-term perspectives:
• We need feedback to tailor the language!
• From both evolution and distributed system communities
• Long-term perspectives:
• Large scale deployment
24
27. Thank you for your attention
Cloud Computing:
From «Revolution» to «Evolution»
1 1,2 1
Sébastien Mosser, Eirik Brandtzæg, Parastoo Mohagheghi,
(1) SINTEF IKT,
(2) University of Oslo
BENEVOL’11,
December 8th 2011