PESOS 2011SUMMARY AND NEXT STEPS

3rd International Workshop on Principles for Engineering Service-Oriented Systems

ICSE 2011Hawaii, USAMay 23-24, 2011

PESOS 2011

2

Keynote: Can Global Software Development and Service-Oriented Architectures Benefit From Each Other?Elisabetta DiNitto, Politécnico di Milano, Italy

How can SOA help GSD/GSE? Distributing/Outsourcing Service Development

○ Communication between teams through service interface definition○ Coordination possibly reduced if composition logic is developed

when services are already available○ Use of standards can help overcome cultural differences○ Service registries can be part of knowledge sharing mechanisms

Service Provision○ This form of adoption extends global service development to global

service engineering due to the runtime aspect○ SLAs play a big role in establishing and maintaining trust

PESOS 2011

3

Keynote: Can Global Software Development and Service-Oriented Architectures Benefit From Each Other? 2 How can SOA help GSD/GSE? (Continued …)

Dynamic Adoption of External Services○ More emphasis on runtime GSE activities such as monitoring and dynamic adaptation○ Services evolve independently of their consumers and vice versa○ Also a problem of trust○ Advantages can arise not only from cooperation but also from competition

Additional perspective: SaaS○ Communication services, CM services, WFM services, deployment services

How can GSD/GSE help SOA? GSD processes

Summary Additional reasons for adoption

○ Availability of suitable and cost-effective services○ Advantages in servicing existing components

New issues○ Runtime aspects, new stakeholders, evolution issues, SLAs, competition

PESOS 2011

4

Presentation: TweetflowsMartin Treiber, TU Vienna and ikangai solutions, Austria

Use of Twitter as underlying infrastructure for services — tweets for all service communication — dedicated Twitter syntaxHad to model find/bind/execute as Twitter primitives

○ Service requests are forwarded to followers that in turn forward requests — use re-tweet structure

○ For binding they use the Twitter addressing mechanism — service could even return a signature to start negotiation and adaptation (future work)

○ Service results/completion are tweeted back to the requestorPossible to delegate or reject a service requestCan also “concatenate” services using Tweetflow pipes

(current work)

PESOS 2011

5

Presentation: Evaluating the Compatibility of Conversational Service InteractionsSam Guinea (Politecnico di Milano, Italy) Motivation is that conversational services impose

a specific interaction protocol (possibly stateful) and data types

Uses Interaction Sequence Charts (ISCs) to describe service interaction with the outside world

Uses compatibility evaluation algorithms to detect protocol and data type mismatches between services

Useful for selecting between candidate services (or execution paths) based on compatibility

PESOS 2011

6

Presentation: Towards Efficient Measuring of Web Services API CoverageWaldemar Hummer, Vienna University of Technology, Austria

Focus is on meaningful API coverage by reduction of value domains

Important pre-condition: Exact definitions are done using XSD Facets

Approach based on domain partitioning (domain partitions are user-defined for value and number of occurrences of each XML node) to reduce value domains

Tool support: TeCOS framework (Test Coverage for Service-Based Systems)

PESOS 2011

7

Presentation: Engineering Multi-Tenant Software-as-a-Service Systems Abhik Roychoudhury (National University of Singapore, Singapore)

Customization of SaaS to accommodate new tenants can be systematized using testing/debugging research

Towards a formal model for SaaS systemsVariant managementCost modelTenant onboarding management

Test only changes introduced by a new tenant

PESOS 2011

8

Presentation: Architecture-Based Reliability Analysis of Web Services in Multilayer EnvironmentsCobra Rahmani (University of Nebraska-Omaha, USA)

Focus is on reliability analysis of web services where some details of underlying layers and configuration settings are taken into account — gray boxes or gray layers

Uses static and dynamic analysis to form the architecture model

Forms a Petri net model based on the architectural model plus a set of parameters related to each layer (transition probability, time spent and failure probability)

PESOS 2011

9

Panel 1

Andrea ZismanService consumer point of view

○ Technique to gather and infer consumer feedback ○ Market regulation — are SLAs enough? What about

trust and reputation?Lack of real-world examples to test ideas — service

interface definitions, case studies, benchmarksComposition adaptation — lots of work but still lots

of problems — for example, adaptation of a set of dependant services as opposed to a single services

PESOS 2011

10

Panel 2

Elisabetta Di NittoLack of application of research results to

industrial problems and/or not reporting those results — industry also often uses research and does not report on results

Processes, techniques, best practices, metrics to accompany existing research

Relationship between SOA and GSD/GSERelationship between SOA and Cloud

Computing

PESOS 2011

11

Panel 3

Liam O’BrienEngaging with industry — still difficult for

researchers — they also unfortunately don’t keep any data

Service orientation in tactical environmentsService orientation for scientists to enable

information sharingService orientation and cloud computing —

e.g. allocation of services to platform for optimal data processing

PESOS 2011

12

Panel 4

Marin LitoiuIntersection of SaaS and cloud computing

○ Elasticity — how do services use only the resources that they need?

○ Automation — full service setup without user intervention (self-serving services)

○ Business models○ Software development life cycle○ Application migration

PESOS 2011

13

Panel Discussion How do companies measure the impact of

services on their operations as well as their revenue? What is the business value of SOA adoption?

How do translate KPIs to IT operational metrics?

The complementary nature of service orientation and cloud is clear, but the challenges and risks are also very clear. How do we make them work together?

PESOS 2011

14

Principles of Engineering Service-Oriented Systems (So Far …) Think globally

Bound instead of control behaviorPlan for diversity

Increase flexibility Reduce complexity Enable agility

PESOS 2011

15

Keynote: Selecting SOA Pilot ProjectsLiam O’Brien (CSIRO, Australia)

Criteria correspond to project characteristics that are analyzed to determine if a project would be a good SOA pilot project

○ Benefits: validates technologies, sustainable investment, pilot preparedness, builds SOA capability and capacity, validates architecture and IT operations

○ Risks: mismatch with SOA maturity, mismatch with governance maturity, riskiness, underlying system not SOA compatible

○ Represented on a Kiveat diagram — easy to visualize if the benefits outweigh the risks

PESOS 2011

16

Presentation: SMaRT: A Workbench for Reporting the Monitorability of Services from SLAsHoward Foster (City University London, UK)

Automation of service monitoring○ Extract SLA terms○ Select matching monitoring components

(reasoners, sensors and effectors)○ Produce an SLA monitorability assessment○ Compose the monitoring components into a

monitoring system configuration (MSC) Eclipse-based tool support

PESOS 2011

17

Presentation: Identifying, Modifying, Creating, and Removing Monitor Rules for Service-Oriented ComputingAndrea Zisman (City University London, UK)

Pattern-based HCI-aware monitor adaptation framework (MADap)Patterns and monitor rules expressed in event

calculus (EC) Semi-instantiated patterns are used to look

for matching rules Patterns are composed of monitor rules and

assumptions about the service-based system Prototype tool implemented

PESOS 2011

18

Presentation: Business Process Performance Prediction on a Tracked Simulation ModelMarin Litoiu (York University, Canada)

Business processes need to adapt to satisfy SLAs — need better KPI prediction techniques

Uses predictive feedback loops to maintain KPIs close to the reference target and predict short-term change than can be use to guide business process adaptation

PESOS 2011

19

Invited Talk: An Architectural Blueprint for Service-Based Science in the CloudIan Gorton, Pacific Northwest National Laboratory, USA

Systems Biology Knowledge Base (Kbase) Collaborative platform Data access platform Plug-and-play tool development platform

Need to account for large amounts of heterogeneity Distributed and diverse data stores, types and access mechanisms Ontologies Workflows Software tools Computation platforms

REST and Hadoop are a good match for the large amounts of parallel data processing that happens in scientific communities

Cloud hosting solutions are a good option for scientists that want to share their applications and data with their community

Metadata is the key to promote sharing and integration

PESOS 2011

20

Principles of Engineering Service-Oriented Systems Think globally

Bound instead of control behavior — runtime simulation, monitoring and adaptation

Plan for diversity Increase flexibility Reduce complexity Enable agility/compositionality — metadata is

key Reduce risk via pilot projects that weigh benefit

against risk

PESOS 2011

21

Next Steps We will add presentations to the PESOS

web site in PDF format (with your permission)

http://www.s-cube-network.eu/pesos-2011/ Workshop summary will be published as

ACM Software Engineering Notes (SEN) We will submit a PESOS 2012 proposal

and count on you for submissions and attendance!

PESOS 2011

22

MESOCA 2011 International Workshop on the

Maintenance and Evolution of Service-Oriented and Cloud-Based Systems

Co-located event with the 27th International Conference on Software Maintenance (ICSM 2011)

When and where? September 26, 2011 in Williamsburg, VA

http://www.sei.cmu.edu/workshops/mesoca/2011/

PESOS 2011

23

Call for ChaptersMigrating Legacy

Applications: Challenges in Service-Oriented Architecture

and Cloud Computing Environments

Publisher: IGI GlobalPublication Date: 2012

IMPORTANT SUBMISSION DATESAbstracts: June 15, 2011Full Chapters: September 15, 2011