This document provides an overview of the Cloud4SOA project and its achievements in cloud computing SLAs. The key points are:
1. Cloud4SOA provides an interoperable framework that supports application developers in deploying and managing applications across multiple heterogeneous cloud platforms.
2. It addresses the need for dynamic SLAs that can adapt over time by defining an SLA management lifecycle that includes negotiation, monitoring, and reaction to violations.
3. The project delivers a monitoring component that collects unified metrics across platforms to provide real-time visibility of application performance against defined SLAs.
2. Outline
1. Cloud4SOA at a glance
• Core Capabilities
• Cloud4SOA Architecture
1. Cloud4SOA SLA achievements and
challenges
• Why we need dynamic SLAs
• SLA Management Life-cycle
1. Real-Time/Unified applications
monitoring
2. In-depth with the technology
3. Future challenges
3. Cloud4SOA at a glance
• The Cloud4SOA project
helps empowers a multi-
cloud paradigm at PaaS
level
• Provides an interoperable
framework for PaaS
developers
Duration: 36 months – (09 / 2010 – 08 / 2013)
Budget: 4.13 M Euros /
Funding: 2.73 M Euros
Research Programme: FP7-ICT-2009-5
Obj1.2
Project coordinator: Atos (Spain)
Partners:
• The system supports Cloud-based application developers with
multiplatform matchmaking, management, monitoring and migration
• Interconnects heterogeneous PaaS offerings across different
providers that share the same technology
app
app
app
4. Matchmaking Management and
Migration
Monitoring
Semantic profiles of:
• User and Applications
• PaaS offerings
Definition of equivalence and compatibility
rules
• Same tools manage different clouds
• Cloud4SOA harmonizes different providers
API
• Use the same interface to deploy and
migrate applications between PaaS providers
• Cloud4SOA defines unified metrics (across
PaaS providers) to monitor application
performances
• Cloud4SOA defines Service Level Agreement
terms and monitors SLA breakages
6. SLA Achievements & Challenges
delivers an enhanced and distributed architecture for business application
governance, monitoring and service level agreement management
– Management Components support the efficient runtime Governance of Application on
Specific PaaS offering.
– Monitoring Component based on unified platform-independent metrics
– Light, dynamic and stand-alone PaaS technological-independent Service Level
Agreement framework to enforce agreements and violation policies at runtime
7. Why we need dynamic SLAs?
From product and established long-term contracts to on-demand based business models –
need for dynamic SLAs that the PaaS ecosystem is not providing:
– SLA’s today are static, long term contracts / almost no state-of-the-art for Dynamic
SLA Management.
Does not support business dynamics and changing customer needs
PaaS providers in the market currently do not have any dynamic SLA capabilities
This makes it difficult to envision an SLA negotiation mechanism, since the
providers won’t be an active party taking part in it.
Max capacity
peak
demand
under -
utilisation
time
UsageUsage
Need to negotiate flexibly between standard and customized SLAs
Need for business-performance related SLA metrics
Users
Need for customer oriented SLA Management
tools which are able to define, monitor, redefine,
correct and report the most important SLA
conditions (life cycle approach)
Users
Need for customer oriented SLA Management
tools which are able to define, monitor, redefine,
correct and report the most important SLA
conditions (life cycle approach)
Providers
Need for SLA tools which enable PaaS
providers to analyse their own
performance to offer reasonable
Business Application SLA
Providers
Need for SLA tools which enable PaaS
providers to analyse their own
performance to offer reasonable
Business Application SLA
8. SLA Management Life-cycle
Template
Creation /
Publication
Cloud
Discovery
Negotiation
Monitoring
and
Evaluation
Reaction
Off-Line time: Cloud Offering
Publication (WS-Agreement)
Design-Time: Search with
Application QoS requirements
Agreement Creation
Cloud Resource /
Application metrics
monitoring –
enforcement of the
signed agreement
Policy to react when the
agreement is not fulfilled PaaS Semantic
Profile
Matchmaking
1
Matchmaking 2
Monitoring &
Evaluation
Violation:
Migration / Rating Design-Time: Develop policies to define
violation when breaches of the
agreement occurs
Limitation of the approach
No commitment of the underling Clouds
C4S (multi-cloud) acts on behalf of the offerings
The result is an agreement Not a real contract
Violation Not Penalties or Discounts
No Charging info in the loop
9. Real-Time and Unified applications monitoring
9
The problem: diversity of metrics and API (dissimilar provider with different metrics with
the same name)
Cloud4SOA’s answer:
•unified metrics, unified interface – overlooking all your deployments at once
•empowers developer to compare, contrast, test
•brings a unique application based user-defined applications policies monitoring
• Application/DataBase RT
• Cloud RT
• Web Container RT
• Application Status
• Memory Consumed
• CPU Consumed
10. In-depth with the technology
WS-Agreement Compatible
Cloud4SOA provides a JAVA REST-full implementation of a SLA
Management layer fully-compatible with the WS-Agreement specification.
•Agreement Template publication and Discovery . The cloud offering is embedded in the
Cloud4SOA Semantic Ontology (OWL)
https://dl.dropboxusercontent.com/u/17326703/c4s_spec/c4sModel.html
•SLA Negotiation. Automate negotiations on behalf of PaaS providers, based on the semantic
description of offerings and the QoS requirements specified by the Application Developer.
•SLA Enforcement. Supervise that all the agreements reached in a SLA agreement are
respected (i.e. measurements are within the thresholds established in SLA agreement for QoS
metrics).
•SLA violation recovery. Whenever the execution of the business application does not satisfy
the SLA (i.e. breaches of the agreement occurs), the most appropriate recovery action (e.g.
warning messages, stop or migration of the application) is suggested based on the policies
defined by the software developer.
The code is released under Apache 2.0 license,
please email me if you want the code: http://demo.cloud4soa.eu/
11. Future challenges
Work on standards (CAMP TWG from OASIS)
Improve performance
SLA Management
– Extend the mechanism to manage complex agreements
– Put in the loop the cloud providers as well (issue so far)
Unified Metrics
– Find out more (unified) monitoring metrics, and categorize them for commercial
and internal users
Useful links:
•Demo Platform: http://demo.cloud4soa.eu/cloud4soa/
•Standalone Matchmaking:
•http://paas-match.cloud4soa.eu/standalone/
•Forum: http://www.forum.cloud4soa.eu/cloud4soa_forum/
•Github: https://github.com/Cloud4SOA/Cloud4SOA
•Please email me @ francesco.dandria@atos.net for detailed design, papers or code.
13. MOdel-Driven Approach for design and
execution of applications on multiple
Clouds
Francesco D’Andria
Francesco.dandria@atos.net
On behalf of Danilo Ardagna (Politecnico di Milano)
danilo.ardagna@polimi.it
14. MODAClouds challenges &
objectives• Focus on needs of Cloud-based Application Developers and
Operators
• Challenges
• Avoid vendor lock-in
• Support risk analysis and management
• Guarantee quality assurance
• To provide methods, a decision support system, an IDE and a
runtime environment to support
• Cost and business model comparison
• High-level design
• Early prototyping
• Semi-automatic code generation
• Automatic (re)deployment
• Monitoring and self-adaptation
of applications on multi-Clouds with guaranteed QoS
14
15. MODAClouds expected
results
15
Developer
CIM
DSS
CPIM
CPSM
Semi-automatic transformation
Automatic deployment
Decision making
New or legacy
applications design
Code development
Design-timeRun-time
Goal: QoS assurance & costs minimization
1
2
4
5
3
Availability
24h/day
Response time <0.4s
CIM
B-1
VM-Large CPU
Instance
A-1
VM-Small CPU
key-valued
DB
Inst. numb. >2
Deploy
on Public
Cloud
CPIM
CPSM
key-valued
DB
A
Reliable
Resource B
High perf.
Resource
C
Large memory
Resource
A-1
VM-Small CPU
B-1
VM-Large CPU
Instance
A-2
PaaS
B-2
PaaS
C
PaaS
Goal: Cost & Risk Analysis
High-level Model-Driven
Application Design
Goal: Verification & validation of requirements
QoS property analysis
Goal: Cloud Independent
Model Representation
spanning across all
abstraction layers
Goal: Cloud Specific
Code Representation
IDE
Monitoring & Data syncronization
Run-time adaptation
Service
Operator
Management
16. 16
Define and SLA and Monitoring architecture supports multi-cloud
application scenarios.
Support the definition of QoS constraints (hard and soft
constraints):
For design time exploration
Triggers for adaptation in case of run-time violations
Enforce agreement at run-time
Automatic triggering of adaptation actions in case of run-time
violations (automatic re-negotiation)
Based on unified run-time monitoring metrics (from IaaS and
PaaS)
1. Reuse outcome from past projects: (cloud4SOA, mOSAIC and other)
2. Still collecting requirements
3. Architecture not yet defined
4. The consortium is willing to use the WS-Agreement specification.
5. First version of the code available only by the next winter
SLA
approach
The Cloud4SOA project helps empowers a multi-cloud paradigm at PaaS level providing an open semantic interoperable framework for PaaS developers and providers, capitalizing on SOA, lightweight semantics and user-centric design and development principles. The system supports Cloud-based application developers with multiplatform matchmaking, management, monitoring and migration by interconnecting heterogeneous PaaS offerings across different providers that share the same technology.
To deal with a SLA in an automatic way (i.e. to have mechanisms that automate the SLA setting up, monitoring and enforcement) the SLA itself has to be expressed in a formalized way using an SLA specification language. WS-Agreement offers a protocol to be followed for the negotiation process and a common understanding (i.e. language) of the objects the negotiation is about. Thus, it enables the automatic creation of SLAs. The outcome of a successful negotiation with WS-Agreement is a SLA with binding character for both parties to deliver a reliable service to the end-user. WS-Agreement is a novel but well-accepted standard for creating and enforcing SLAs in distributed environments as well as monitoring their resources properties.
Talk to Jim and me about Beta participation, we’d like you to join. Want to learn more about PaaS and its potential? We have a white paper published by Atos’ Science Committee, where Cloud4SOA contributed. Also, join our LinkedIn Group, as we roll out this release we’ll use that channel to keep you informed of Cloud4SOA news.