An Open Grid Services Architecture for Mobile Network
Download
Report
Transcript An Open Grid Services Architecture for Mobile Network
Cracow Grid Workshop ‘06
17 October 2006
Execution Management and
SLA Enforcement in Akogrimo
Antonios Litke, Kleopatra Konstanteli, Vassiliki
Andronikou, Sotirios Chatzis, and Theodora Varvarigou
e-mail: [email protected]
Presentation Outline
Cracow Grid Workshop ‘06
Architectural overview of Akogrimo
The need for SLAs
Component descriptions and interactions
Implementation issues
Related work
Conclusions
Questions/answers
17 October 2006
OGSA-based Layered Architecture
Cracow Grid Workshop ‘06
17 October 2006
An OGSA compliant architecture aims to
Manage
resources
heterogeneous platforms
across
distributed
Deliver seamless access control and quality of
service
Provide
a
common
management solutions
base
for
Define open, published interfaces
autonomic
Akogrimo Architecture
Cracow Grid Workshop ‘06
Akogrimo
Applications / Scenarios
OGSA
17 October 2006
OGSA specific services
Execution Management
Services
Information
Services
Data Management
Services
Service Level
Agreement Enforcement
Security Services
WSRF – Based services
WSRF
WSRF – Enabled
Resources
Web Services
WS – Resource
LifeTime Management
WS – Resource
Properties Management
WS – BaseNotifications
WS – Addressing
Equipment
Communications
IP Transport – Mobile IP
Logical Resources
QoS adaptation
Physical Resources
(Grid resources)
Ethernet
802.11
UMTS
L2
Akogrimo Architecture
Cracow Grid Workshop ‘06
17 October 2006
•Physical and logical resources layer
Resources comprise each and every capability of the Grid
•Physical resources, which include servers, storage, and
network and Logical resources.
•Logical resources are above the physical and provide
additional functionality by virtualizing and aggregating the
resources in the physical layer.
•In Akogrimo communication
incorporated in logical resources
resources
are
also
The need for management of SLA
Cracow Grid Workshop ‘06
17 October 2006
Grids and mobile Grids are dynamic environments subject to
unpredictable changes:
system
or network failures, system performance
degradation, removal of machines, variations in the cost
of resources,
Execution Management components should take care of
conformance to the contractual terms of SLAs
EMS system monitors and manages the execution of the
job until its completion
In case of violations, it takes actions to dynamically
rectify them in such a way as to meet the terms defined
in the related SLA
Grid Infrastructure Services
Cracow Grid Workshop ‘06
17 October 2006
SLA contractual terms that
influence the execution of jobs
us
St Su ines
atu bs s
s n cr i b r e s
oti e our
ce
fic
ati
on
especially
•Monitoring:
status of the execution, availability of
services, and of gathering the relevant
information
EMS
Create resource
Activate/Destroy
Sta
rt/
Sto
p
mo
nito
ri n g
SLAController
Subscribe
Monitoring
notification
te
b
Cr
•SLA Enforcement:
Violation
detected
Instantiating and managing tasks
SLA-Decisor
Register
Retrieve QoS
parameters
•Execution Management Services:
Business
Application
Service
SLA Access
ea
•Addresses the performance issues while
conforming to the determined Service Level
Agreement (SLA).
Monitoring
Example of QoS Parameters in SLAs
Cracow Grid Workshop ‘06
<QoSParams>
<QoSItems>
<QoSItem>
<param>cpuSpeed</param>
<paramValue>3.5</paramValue>
<paramType>MHz</paramType>
<threshold>40%</threshold>
</QoSItem>
<QoSItem>
<param>diskSpace</param>
<paramValue>1</paramValue>
<paramType>GB</paramType>
<threshold>20%</threshold>
</QoSItem>
<QoSItem>
<param>memory</param>
<paramValue>1</paramValue>
<paramType>GB</paramType>
<threshold>50%</threshold>
</QoSItem>
<QoSItem>
<param>networkBandwidth</param>
<paramValue>GOLD</paramValue>
<paramType></paramType>
<threshold></paramValue>
</QoSItem>
</QoSItems>
</QoSParams>
17 October 2006
Discovery and Reservation phase
Cracow Grid Workshop ‘06
Service
17 October 2006
Interaction
Interface
SLA-Access
Step 2
ObjQoS getQoSParameters(String slaID)
SLA-Discovery
Step 3
ArrayList findSuitableResources(objQoS objQoSData)
Business service
Step 5
EndpointReferenceType create()
QoS Broker
Step 6
Boolean setQoS(EndpointReferenceType serviceEPR, integer networkBundleType)
SLA-Controller
Step 8
EndpointReferenceType create()
Step 9
Boolean registerToServiceDecisor(EndpointReferenceType slaControllerEPR,
EndpointReferenceType serviceEPR)
Client
Discovery and Reservation MSC
Cracow Grid Workshop ‘06
EMS
17 October 2006
SLA-Access
Business
service
SLA-Discovery
QoS Broker
SLA-Controller
SLA-Decisor
1: Specify slaID and
userID, submit request
for discovery and
reservation of resources
2: Retrieve QoS
parameters from SLA
3: Look for resources
R
4: Match? Create a
service reservation
resource and add related
info (slaID, userID, QoS)
Reservation
Resource
5: Create an
business service
resource and manage
its lifecycle
6: Add the EPR of
the e-Health resource
to the reservation
7: Perform network
resources reservation
8: Create a SLA-Controller
resource
11: Return the EPR of
the reservation resource
10: Add the EPR of the SLAController resource to the
reservation resource
9: Register to the
SLA-Decisor
Execution and Monitoring phase
Cracow Grid Workshop ‘06
Service
SLA-Controller
17 October 2006
Interaction
Interface
Step 4
Boolean activateServiceController(EndpointReferenceType serviceEPR, objQoS objQoSData,
String slaID)
Step 11
Boolean receiveInformationByMonitoring(EndpointReferenceType serviceEPR, objQoS
objQoSData)
Step 17, 24
Void destroy()
SLA-Decisor
Step 12
Boolean setViolation( EndpointReferenceType serviceEPR, Violation violation)
Monitoring
Step 5
Boolean startMonitoring( EndpointReferenceType serviceEPR, objQoS objQoSData, String
slaID )
Step 14, 21
Metering
Business
Boolean stopMonitoring( EndpointReferenceType serviceEPR)
Step 6
org.oasis.wsn.SubscribeResponse subscribe(org.oasis.wsn.Subscribe subscribeRequest)
Step 7
Boolean startMetering( EndpointReferenceType serviceEPR)
Step 16, 23
Boolean stopMetering( EndpointReferenceType serviceEPR)
Step 8
Not standard (depends on the e-Health service)
Execution and Monitoring MSC
Client
Cracow Grid Workshop ‘06
EMS
1: Specify EPR of
reservation resource,
input for the service
and submit request
R
SLA-Controller
SLA-Desicor
SLA-Metering
2. Find reservation
resource
Business
Resource
R
Reservation
Resource
3. Retrieve info
4: Activate SLAController resource
5: Invoke Monitoring Service
6: Subscribe for notifications
7: Invoke Metering Service
8: Start execution of Business Resource
9: Loop (Monitor
execution)
11: Notify SLAController about
the change
21: Go back
to step 3
and repeat
sequence
10: Significant change? Notify
SLA-Monitoring
12: Violation? Notify SLA-Desicor
13: Decide on
appropriate action
14: Reallocation needed? Notify EMS
15: Stop Monitoring service
16: Stop Metering Service
17: Destroy SLAController resource
18: Destroy Business Resource
19: Find new resources
20: Update Reservation Resource
22:Done? Stop Monitoring service
23: Stop Metering Service
24: Destroy SLAController resource
25: Destroy Business Resource
26: Destroy Reservation Resource
27: Return the output of
the business service
execution
17 October 2006
R
SLA-Monitoring
Implementation issues
Cracow Grid Workshop ‘06
17 October 2006
Globus Toolkit 4, and especially WS-GRAM and
MDS4
SLA contracts management are developed using
the WSRF.NET platform
Compelling
implementation
challenge to see whether:
and
technological
•The 2 main grid services development tools,
implement the WS-related specifications in a
transparent and interoperable way
Related work
Cracow Grid Workshop ‘06
17 October 2006
GRIA (Grid Resources for Industrial Applications)
•designed and developed the GRIA middleware based on Web Services
•supports the confirmation of a service offer through establishment of
SLA and extensions of existing SLAs
GRASP (Grid-based Application Service Provision)
•SLA management subsystem including the service provision negotiation
based on QoS criteria and monitoring of the feasibility of the contract
Unigrids (Uniform Interface to Grid Services)
•SLA framework and cross-Grid brokering services in order to support
Grid economics
•integrating a Web Services Agreement-based resource management
framework into the Unicore Grid middleware
...and what is special in Akogrimo?
Cracow Grid Workshop ‘06
1) SLA
parameters
parameters
17 October 2006
do
include
network
QoS
2) Negotiation, reservation and execution are made
on this basis also
3) Utilization of both toolkits: GT4 and WSRF.NET
4) Decision on appropriate action when violation and
estimation of the significance of the violation
through a dedicated component
Conclusion
Cracow Grid Workshop ‘06
17 October 2006
•We presented an architecture to manage execution and
enforce SLA in OGSA based grids and mobile Grids
Discovery and reservation phase
Execution and monitoring phase
•The proposed design has been implemented in GT4 and
WSRF.NET
•Advancing attributes of SLA mangement with use of network
resources
•For future research: To define an economic model based on
how these violations affect the efficiency of Grids
Cracow Grid Workshop ‘06
17 October 2006
Thank you!
Questions?
Cracow Grid Workshop ‘06
17 October 2006
Backup slides
Akogrimo Architecture
Cracow Grid Workshop ‘06
17 October 2006
•Web services layer
All Grid resources are modelled as services
Web Services Resource Framework (WSRF) defines a family of
specifications
services
for
accessing
stateful
resources
using
Web
•OGSA architected grid services layer
Overall grid management functionality
•Grid applications layer
High – level applications a client might require
For Akogrimo we have deployed an E – Health application
offering a suite of E – Health services.
Grid Infrastructure Services
Layer
Cracow Grid Workshop ‘06
17 October 2006
Network QoS
Cracow Grid Workshop ‘06
17 October 2006
•The Mobile Network Layer implements end-to-end Quality of Service,
for all types of mobility
Make an effective use of network resources
Assure that a user can utilize the services he is entitled to without
disruptions
•The QoS implementation
Allows fine-grained QoS control at the access networks
Aggregates different flows with the same QoS requirements in the
core network
•The network supports
well defined QoS bundles
Bundle 1
Mix audio + data
10 – Interactive
100 – Data
1 – Priority
1 – Signalling
250 – Best Effort
Bundle 2
High data + video
20 – Interactive
1000 – Data
200 – Priority
1 – Signalling
Bundle 3
Mostly voice
10 – Interactive
1 – Priority
1 – Signalling
250 – Best Effort
Application Services Layer
Cracow Grid Workshop ‘06
17 October 2006
•Each application service requested by a client is modeled as a
business process
•Workflows represent the automation of the business process
Each workflow coordinates and manages component services or
entities involved into the automation of business process
•A Virtual Organization (VO) provides services and the means
to manage and coordinate Business Processes
•Implementation of a Business Process implies
Creation of an Operational Virtual Organisation (OpVO) out of a
base VO.
MDVO example focused on AAA
Cracow Grid Workshop ‘06
17 October 2006