• No results found

Information Federation in Grid Information Services

N/A
N/A
Protected

Academic year: 2020

Share "Information Federation in Grid Information Services"

Copied!
36
0
0

Loading.... (view fulltext now)

Full text

(1)

Information Federation in Grid

Information Services

Mehmet S. Aktas

(2)

Talk Outline

§

Use Cases and Challenges

§

Research Issues

§

Architecture

Ø

Hybrid Grid Information Service

§

Performance Evaluation

§

Conclusions

(3)

Introduction

§

Grid Information Services in Service Oriented Architectures

Ø

1) Large scale relatively static metadata as in catalog of all the

world’s services

• Interaction-independent, slowly-varying metadata

Ø

2) Small scale highly dynamic metadata as in dynamic workflows

for sensor integration and collaboration

• Interaction-dependent, dynamic metadata • Dynamic Grid/Web Service Collections*

– Dynamically assembled relatively small number of services (sub-grid) – Gathered at any one time to support a specific task

(4)

Motivating Use Cases

§

Geophysical Data Grids - CGL

Ø Service Oriented Architecture for Geographical Information Systems

Supporting Real Time Data Grids

§

Pattern Informatics (PI) - UC Davis

Ø Earthquake forecasting code developed by Prof. John Rundle (UC

Davis) and collaborators, uses seismic archives.

§

Interdependent Energy Infrastructure Simulation System

(IEISS) - LANL

Ø Models infrastructure networks (e.g. electric power systems and

natural gas pipelines) and simulates their physical behavior, interdependencies between systems.

§

eSports System - CGL

Ø Annotative collaboration application. Supports archive, replay,

(5)
(6)

Background

§

Specifications for interaction-independent metadata

Ø UDDI Specification

Ø Glue Specification

Ø EbXML Specification

Ø Web Registry Service Specification

§

Specifications for interaction-dependent metadata

Ø Point-to-point approach

• Web Service Resource Framework (WSRF) Specification

Ø Third-party approach

(7)

Challenges

§

Standardization and Unification Issues

Ø Customized Grid Information Services

Ø Fat clients

§

Performance and Centralization Issues

Ø Low performance

Ø Low fault tolerance

§

UDDI Specification Issues

Ø Lack of up-to-date, metadata-oriented registry Ø Lack of domain-specific metadata management

§

WS-Context Specification Issues

(8)

Research Issues I

§

Unification

Ø How to combine different information services?

§

Federation

Ø How to federate different information services?

§

Flexibility

Ø How to accommodate broad range of specific application domains?

§

Interoperability

Ø How to facilitate connection with wide range of information service

(9)

Research Issues II

§

Performance

Ø

How to provide efficient information management strategies?

• high-performance, scalable in-memory storage • efficient request distribution

• adaptation to instantaneous client-demand changes

§

Fault-tolerance

Ø

How to provide efficient replica-content placement strategies?

§

Consistency

(10)

Hybrid Grid Information Service

§

Unification

§

Federation

ØUnified Schema ØQuery/Publish API

§

Flexibility

§

Interoperability

ØExtended UDDI ØWS-Context ØGlue

Ø…

Hybrid Grid Information Service

§

Unification

§

Federation

ØUnified Schema ØQuery/Publish API

§

Flexibility

§

Interoperability

ØExtended UDDI ØWS-Context ØGlue

(11)

UDDI instance

WS-Context instance

(12)
(13)

Support for interaction-independent

metadata:

Extended UDDI Service

§

There are other extensions of UDDI

§

Supports different types of metadata

ØUser-defined metadata ØFunctional metadata

§

Enables advanced query capabilities

ØGeo-spatial, metadata-oriented, domain-independent queries

§

Provides additional capabilities

ØUp-to-date service registry information (leasing) ØDynamic aggregation of capabilities of services

e.g. geospatial capabilities

[GGF16-Semantic Grid Workshop] Web Service Information Systems and Applications http://www.semanticgrid.org/OGF/ggf16/papers/GGF16SemGrid-CGL.pdf

[SKG06 – IEEE Proceedings] XML Metadata Services

(14)

Support for interaction-dependent

metadata:

WS-Context Service

§

OASIS Standard

§

Context Manager Service

Ø Data model and communication protocol

§

Supports Dynamic Web Service Collections

Ø Distributed state based systems e.g. workflow-style grids Ø Session metadata management

e.g. real-time replay and session-failure recovery capabilities

§

Provides various capabilities

Ø Notification capability

Ø Up-to-date metadata registry (leasing)

[SKG05 – IEEE Proceedings] Information Services for Dynamically Assembled Semantic Grids http://grids.ucs.indiana.edu/ptliupages/publications/skg05-56-maktas-ieee-version.pdf

(15)

Support for federated service

metadata:

Information Federation

Support for federated service

metadata:

Information Federation

§

Federating Grid Information Services

ØUnified Schema and communication protocol ØExtended UDDI, WS-Context and Glue Schemas

§

Approach taken for Unified Schema [Schema Integration]

ØSchema Matching

Identify overlapping information in given two Schemas: S1 and S2 ØSchema Merging

Use the identified overlapping information to guide merge of S1 and S2

§

Communication protocol

ØPublish: save_ (create, update), delete_ e.g. save_service, delete_service

ØInquiry: find_ , get_

(16)

Schema Matching: Identifying Matching Concepts

serviceAttributeEntity: Information about metadata associated to services

Site

Service ComputingElement StorageElement

site:information about a site where services, computing elements and storage elements are aggregated

ServiceData

service:all information about a Service

ExtUDDI.businessEntity 1:N GLUE.site

Extended UDDI

GLUE

(17)

metadata:information about metadata associated to service

bindingTemplate:Technical information about a service point

tModel:Description of Specifications for services or taxonomies

publisherAssertions:

Defines relationships between two business entities

computingElement:all info. required to manage

computing resources

storageElement:all information required to manage storage resources

businessEntity:information about the party who publishes information about entities

service:all information about a service

site:all information about a concept to aggregate services and resources

site contains one to n computing element

has references to

site contains one to n services

site contains one to n storage element business contains

one to n services has references to

service contains one to n metadata service contains one to n

technical information

business contains one to n site

Schema Merging: Unifying Schemas

Unified Schema GLUE

(18)

Key Design Features

§

In-Memory storage

Ø

High performance metadata access/storage

§

Access distribution

Ø

Redirecting client request to an appropriate replica server

§

Replica content placement for performance

Ø

Dynamic replication

Moving/replicating metadata to where they are demanded.

§

Replica content placement for fault-tolerance

Ø

Permanent replication

Replicating data on an appropriate replica server

§

Consistency enforcement

(19)

In-Memory Storage

§

Light-weight implementation of JavaSpaces

Ø

Data sharing, associative lookup

§

Integrated in-memory storage capability

Ø

Ex: UDDI-type, WS-Context-type

Ø

Today’s servers are capable of holding such small size metadata

in memory.

§

Persistency

Ø Newly-inserted/updated metadata is backed-up into appropriate

information service back-end.

Ø If the physical memory wiped out, at the bootstrap,

(20)
(21)
(22)
(23)
(24)

Access Distribution and Dynamic Replication

§

Broadcast-based request dissemination

Ø

Pub-sub system for message broadcast

Ø

Requests are broadcast only to those servers that can answer

Ø

No need to keep track of metadata locations

§

Replica-content placement for performance

Ø

Popular copies are moved/replicated where they are demanded

Ø

Dynamic migration/replication algorithm

*

(25)

Access Distribution Experimen

Benchmark Methodology

T

1 T2 T3

Time = T1 + T2 + T3

One-broker case

(26)

Experiment Results

(27)

Experiment Results

(28)

T

1 T2 T3

Time = T1 + T2 + T3

Dynamic Replication Performance Experimen

Benchmark Methodology

Simulation parameters

message size / message rate 2.7 Kbytes / 10 msg/sec replication decision frequency every 100 seconds

(29)

§

The decrease in average latency shows that the algorithm

manages to move replica copies to where they are demanded.

(30)

Replication and Consistency

§

Permanent replication for fault tolerance

Ø Each node keeps information about other servers Ø Replica Server(s) Selection

• Load and proximity metrics

• Selection algorithm by Rabinovich et al

Ø Unicast-based replica-content placement

§

Primary-copy approach

Ø Updates are unicast to primary-copy

Ø Updates are broadcast by the primary-copy holder to

• a) permanent-copy holding servers

(31)

Fault-tolerance Experiment

Benchmark Methodology

T

1 T2 T3

Time = T1 + T2 + T3

(32)

Experiment Results

(33)

Consistency Enforcement Experimen

Benchmark Methodology

T

1 T2 T3

Time = T1 + T2 + T3

One-broker case

(34)

Experiment Results

§

Overhead of consistency enforcement is few milliseconds.

(35)

Contributions

§

Systems Research

Ø

Hybrid Grid Information Service Architecture

vUnification, Federation and Interoperability of grid information services

vStrategies for high-performance, scalable in-memory storage vStrategies for efficient distribution, replica-content placement, consistency enforcement by utilizing pub-sub based messaging schemes

vSelf-adaptation to changing-client demands

vExtensions to semantics of UDDI and WS-Context Web Service Specifications

Ø

Detailed evaluation of the system components and algorithms

§

Systems Software

Ø

An implementation of Extended UDDI Specification

vGeographical Information Systems-specific, metadata-oriented

Ø

An implementation of WS-Context Specification

(36)

Future Research Directions

§

Use the proposed approach to solve OGF Grid

Interoperation Now (GIN) problem for information

services

§

Investigate an information security mechanism for

the decentralized Hybrid Service

Ø

Example motivating application case: Pattern Informatics

application

§

Applying Hybrid Service to broader range of

application use cases

References

Related documents

Diagnostic pertaining to literacy and essential skills Scott Murray, President, DataAngel Policy Research Evidence from research on return on training investment Allan

We have Desigual Brand Stores in Barcelona, Madrid, Paris, London, New York, Amsterdam, Berlin, Lisbon, Bahrain, Dubai, Saudi Arabia, Singapore, Athens, Kuwait, Moscow, Cairo....

The building located in Niksar, also known as Çöre¤i Büyük Mosque, is an ‹lhanl› monu- ment, built in the 14th century during the reign of Ebu Said Bahad›r Han. The square

• Create purpose for each employee such that they are motivated to refer • Arm yourself with metrics such as specific source hires within your ERP • Refresh your employee

My father´s name is Robert, he Works as a police- man so he has to gets-up My father´s name is Robert, he Works as a police- man so he has to gets-up very early every day, he is

Even minor deceptive methods, in social-psychological experiments, were ruled out unless given special approval by an institutional review board (I.R.B.) - a watchdog

The specific objectives included to: assess the relationship between employee salary and employee turnover in a manufacturing firm in South East, Nigeria and determine the

Given the challenges and opportunities of peer learning, years of experience, the present researchers have had as educators pointed to the different viewpoints