Internet of Services

fortnecessityusefulDéveloppement de logiciels

14 déc. 2013 (il y a 3 années et 8 mois)

71 vue(s)

From

Service
Markets

to

Service
Economies



An
infrastructure

for

protocol
-
generic

SLA
negotiations

Dagstuhl

Seminar on Service Level Agreements in
Grids

22.


27. 03. 2009

Sebastian Hudert

2

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Outline

Vision: Service
Economies

Internet of Services

Service
Markets

Service
Economies

Research Goal

A
protocol
-
generic

SLA
negotiation

infrastructure

Open
Questions


3

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Internet of Services (1)

Grid Computing:

Dynamic
sharing

and

coordination

of
resources

not
subject

to

central

control
,
delivering


nontrivial

qualities

of
service
“ (Foster2002)


Cloud

Computing:

Dynamic
provision

of inter
-
connected

and

virtualized

computers

on
demand
(Buyya2008)


Enterprise SOA

Variety

of
single

applications

or

complex

workflow

services

are

offered

over

an
intra
-

and

inter
-
company

network

Elaborate
business

applications

are

implemented

as

workflows

over

these

simple
and

complex

services


4

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Internet of Services (2)

Common Vision:
Internet of Services




Service Orientation

Distributed Service Systems

Services
as

basic

building

blocks

for

more

complex

Applications
/Workflows




Decentral

Systems

(
Almost
)
no

hierarchical

control

Open,
decentrally

managed

systems


cross
-
organizational

setting

Value
-
added

Services
based

on Service Orchestration

5

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Outline

Vision: Service
Economies

Internet of Services

Service
Markets

Service
Economies

Research Goal

A
protocol
-
generic

SLA
negotiation

infrastructure

Open
Questions


6

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Commercial
service
-
based

Systems:

Towards

Service
Markets

Service
Markets

IoS

application

in a
commercial

context



Need
for

QoS

guarantees

Cross
-
organisational
setting



Need
for

decentral

coordination


Need
for

a
shared

communication

language


Idea
:
usage

of
socio
-
economic

mechanisms

to

implement

efficient

service
-
based

system

across

organisational
boundaries


Service Level Agreements


electronic
Negotiations




8

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Efficient

Ressource
Allocation
:

Service
Markets

Research
questions

concerning

Service
Markets
:


Efficiency of
economic

self
-
organisation

approaches


for

coordinating

Service
Markets



CATNETS




Development of
methods

and

tools

for

an
efficient
,
market
-
based



resource

allocation

system

aiming

at

the

end
user



SORMA



Investigation of
reputation

and

trust

models

for

a
more


efficient

partner

selection

in Service
Markets



eRep


9

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Outline

Vision: Service
Economies

Internet of Services

Service
Markets

Service
Economies

Research Goal

A
protocol
-
generic

SLA
negotiation

infrastructure

Open
Questions


10

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Next
Step
:

Service
Economies

(1)

Economic

Theory
:

Different
market

situations

demand

different
negotiation

protocols

to

reach

the

highest
-
possible

overall

efficiency
.


Computer Science:

Current

systems

are

generally

based

on
the

same
communication

infrastructure

(Web Services, SOAP etc.)

However
, on a
higher

level

individual
markets

mostly

employ

only

one

particular

protocol

(
aiming

at

the

respective

setting
?).


But
what

if

service

consumers

require

services

that

are

not
available

on
the

market

/
the

sytem

they

are

implemented

for
?

e. g.
for

innovative
and

dynamic

business

processes

11

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Next
Step
:

Service
Economies

(2)

Grid

Computing

Cloud

Computing
(
resource

side
)

Applications

in
the

Cloud

Enterprise SOA

Infrastructure

IP
-
based Intra
-
/Internet

IP
-
based Intra
-
/Internet

IP
-
based Intra
-
/Internet

IP
-
based Intra
-
/Internet

Communication

SOAP, http

SOAP, http

SOAP, http

SOAP, http

Paradigm

(atomic) digital services

(
atomic
) digital
services

(atomic) digital services

(atomic) digital services

Interfaces

Web Services (WSDL)

Web Services (WSDL)

Web Services (WSDL)

Web Services (WSDL)

Transparency

up to actual resource
services

up to actual resource
serivces

up to virtualization layer

up to application
services

Invocation


predictable
, parallel

on demand, parallel
and/or sequential

predictable within the
workflow (if present),
parallel and/or
sequential (according to
WF)

predictable within the
workflow, parallel and/or
sequential (according to
WF)

Point
of

control

scheduler
/
job

submitter

virtualization layer

user/workflow scheduler

user/workflow scheduler
and/or moving

System
Configuration

few to many SC,

many resource services
(SP)

few virtualization
services (SC) ,

many resource services

many SC, few services

many SC, few services

Service
Complexity

standardized commodity
services

standardized commodity
services

standardized and/or
complex services

standardized and/or
complex services

CDAs or Reverse
Auctions (depending on
number of consumers)

Reverse
Auctions

Auctions or Bargaining
(depending on service
complexity)

Auctions or Bargaining
(depending on service
complexity)

12

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Next Step:

Service Economies (3)

Service Market II

Service Market III

Service Market I

Service Economy

13

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Outline

Vision: Service
Economies

Internet of Services

Service
Markets

Service
Economies

Research Goal

A
protocol
-
generic

SLA
negotiation

infrastructure

Open
Questions


14

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Research Goal (1)

Research
Question
:



Is
it

possible

to

increase

the

overall

efficiency

of an SLA
management

layer

in
current

IoS

settings

by

introducing

an
agent
-
based
,
protocol
-
generic

negotiation

infrastructure

and

thus

an Economy of Services?


To what extent can such an infrastructure increase flexibility of the overall
system?

How robust is such an infrastructure?


15

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Research Goal (2)

SLA
-
based

QoS



Management





Need
for

IoS
/SE


Infrastructures


directly

supporting



the

overall

Service


Lifecycle


Focus:

(Discovery
and
)

Negotiation

Phase

16

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Outline

Vision: Service
Economies

Internet of Services

Service
Markets

Service
Economies

Research Goal

A
protocol
-
generic

SLA
negotiation

infrastructure

Open
Questions


17

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Service Economy Infrastructure:

Architecture

Design

Abstract Design
Idea
:


18

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Management vs. Service Layer

Negotiation

components

(Management
Agents
)
are

part

of a
general

Service Management
Overlay

Network

Analogy
: Agreement
and

Service Layer in WS
-
Agreement














WS
-
Agreement
Architecture

(Andrieux2005)

19

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Infrastructure Components

Negotiation

Protocol Description Language


Set of
protocol

primitives
supporting

the

discovery

of
the

respective

SLA
templates

and

negotiation

protocol

descriptions

a flexible
negotiation

phase

(
conducted

according

to

the

protocol

description

discovered
)




20

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Protocol Description
-

Datamodel

22

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Protocol Primitives

Idea
:

Role
-
based

interface

definitions

for


Service Providers

Consumers

and


Intermediaries (
implementing

a
system

of
lookup

and
/
or

auction

servers
)


supporting

the


Discovery
Process

(
requesting

or

proposing

of,
pub
/
sub

functionality

for

protocol

descriptions

etc.)

Negotiation

Process

(
posting

or

accepting
/
rejecting

bids

etc.)


Actual

Negotiation

Process

will
follow

the

specifications

stated

in
the

discovered

protocol

description

document
.

24

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Outline

Vision: Service
Economies

Internet of Services

Service
Markets

Service
Economies

Research Goal

A
protocol
-
generic

SLA
negotiation

infrastructure

Open
Questions


25

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert

Open
Questions

Which

market

configurations

(in
terms

of
service

consumers
,
providers

and

intermediaries)
are

present

in
current

Service
Markets
?


How

do
currently

used

SLAs
look

like
?
To

what

extend

can

economic

theory

prescribe

fitting

negotiation

protocols

for

these

different

products
“?


Which

metrics

are

applicable

for

such
settings
?
How

can

economic

efficiency

and

infrastructural

flexibility

be

measured
?




26

Doktorandenseminar WS 07/08

03. 12. 2007

Sebastian Hudert





Thank

you

for

your

attention
!


Any

questions
?






Sebastian Hudert

Department of Information Systems Management

University of Bayreuth

sebastian.hudert@uni
-
bayreuth.de