Title: Grid Computing Tutorial
1Grid Computing Tutorial
- Ed Green
- Lecturer IST
- The Pennsylvania State University
- The Abington College
- 215-881-7332
- exg13_at_psu.edu
- www.personal.psu.edu/exg13
2Grid Technology Abstract
- Emerging new field
- Beyond distributed computing
- Focus
- Large-scale resource sharing applications
- High-performance orientation
- Requires
- Flexible, secure, coordinated resource sharing
- Involves dynamic collection of
- Individuals
- Institutions
- Resources
- Characterized by unique authentication,
authorization, resource access, and resource
discovery
The virtual organization
3Introduction and Background
- Term grid originated mid 1990s
- Proposed infrastructure for science and
engineering - Expanded to include broadest technology spectrum
- From advanced networking
- To artificial intelligence
- And everything in between
- Addresses real and specific problem space
- Distinct and separate from popular technology
trends - Internet
- Enterprise computing
- Distributed computing
- Peer-to-peer computing
- Symbiotic opportunities when popular technologies
grow into the grid problem space
4Grid Problem Space
- Coordinated resource sharing and problem solving
in dynamic, multi-institutional, virtual
organizations - Essential needs
- Highly flexible sharing relationships ranging
- From client-server
- To peer-to-peer
- Sophisticated and precise levels of control over
use of shared resources - Fine-grained and multi-stakeholder
- Access control
- Delegation
- Application of local and global policies
- Sharing of resources
- From programs, files, and data
- To computers, sensors, and networks
- Diverse usage modes
- From single-user to multi-user
- From performance sensitive to cost-sensitive
- Quality of service
- Scheduling
- Co-allocation
Not addressed by current generation of
distributed computing technologies
5What Grid Technology Offers
- Security solutions that support management
credentials and policies across multiple
enterprises - Resource management services and protocols to
support - Secure remote access to computing and data
resources - Co-allocation of multiple resources
- Information query protocols and services that
provide configuration and status information
about - Resources
- Organizations
- Services
- Data management services that locate and
transport datasets between storage systems and
applications
Compliment existing distributed computing
technologiesrather than competing with them!
6Grid Technology Placement a Perspective
- Virtual organizations
- Set of collaborating enterprises
- Viewed as a single logical entity
- Leverage collaborator
- Processes
- Policies
- Systems
- Resources
- Single-enterprise viewpoint (s)
- Collaboration among diverse business units
- Merger/acquisition/divestiture ramifications
- Cooperative processing among less-than-compatible
systems - Multi-enterprise viewpoint (m)
- Collaboration among diverse enterprises
- msn where n is the number of enterprises
- Global considerations
7Virtual Organizations
- Collaboration to achieve a common goal
- An enterprise can participate in multiple virtual
organizations - Domain-relevant
- Market-centric
- Industry-oriented
- Problem-centric
- Opportunity-centric
- Economics-driven
- Dynamic over time
- Resource sharing is managed
- Need to know accessibility
- Conditional availability who, what, when, how
- Discovery mechanism required to characterize the
state of relationships at some particular point
in time - Peer-to-peer considerations
- Providers and consumers
- Subset relationships
- Single resource, multiple sharing opportunities
8Concept of Grid Architecture
- Grid architectures require establishment of
sharing relationships among potential
participants - Central issue ? interoperability ? protocols
- Grid architecture is a protocol architecture
- Mechanisms for users and resources to negotiate,
establish, manage, and exploit sharing
relationships - Standards-based open architecture
- Facilitates extensibility, interoperability,
portability, and code sharing - Standard protocols enable definition of standard
services that provide enhanced capabilities - Application Programming Interfaces (API)
- Software Development Kits (SDK)
9Importance of Interoperability
- Need to initiate sharing relationships among
arbitrary partners - Need to accommodate new partners dynamically
across different computing environments - Hardware
- Software
- Need to promote multilateral sharing arrangements
- Avoid bilateral resource sharing
- Ensure availability of sharing mechanisms in a
dynamic partnership environment
10Importance of Protocols
- Protocol definition specifies
- How distributed system elements interact with
each other to achieve a specified behavior - Structure of information during interaction
- Virtual organizations compliment existing
enterprises/institutions - Sharing mechanisms must avoid substantial changes
to local policies - Sharing must preserve individual (institution)
control of (their) resources - Protocols
- Govern the interaction between components
- Do not govern implementation of components
- Without standard protocols, interoperability
requires - Single implementation at the API level or
- Having every implementation know details of every
other
11Grid Architecture Description
Application
Application
Collective
Internet Protocol Architecture
Resource
Grid Protocol Architecture
Transport
Connectivity
Internet
Fabric
Link
A relationship exists between the Grid
Protocol Architecture and the Internet Protocol
Architecture.
12Fabric Layer Local Control Interface
- Provides resources to mediate shared access to
system facilities by Grid protocols - Physical system facilities require external
protocols computational components, storage
systems, catalogs, network and/or sensors - Logical system facilities require internal
protocols distributed file system, computer
cluster, and/or distributed computer cluster - Implements local, resource-specific operations on
specific (logical or physical) resources as the
result of higher-level sharing operations - Interdependence between fabric-layer functions
and sharing operations - Tightly coupled
13Delivering Functionality
Application
Co-reservation Service API SDK
Co-reservation Protocol
Collective Layer
Co-reservation Service
Co-Allocation API SDK
Resource Mgmt API SDK
Resource Layer
Resource Management Protocol
Network Resource
Network Resource
Compute Resource
Compute Resource
Fabric Layer
- Can be combined in a variety of ways to deliver
functionality to applications
14Fabric Layer Local Control Interface
- Minimum implementation
- Enquiry mechanisms that permit discovery of
resource structure, state, capabilities - Resource management mechanism that provide
control of delivered quality of service - Capabilities
- Computational resources starting, monitoring,
and controlling the execution of programs - Storage resources getting and putting of files
- Network resources managing network transfers
- Code repositories managing versioned source and
object code - Catalogs implementing catalog query and update
capabilities
15Connectivity Communication and Authentication
Protocols
- Communication protocols enable exchange of data
between Fabric Layer resources - Communications includes
- Transport
- Routing
- Naming
- Authentication protocols
- Build on communications services
- Provide cryptographically secure mechanisms for
identity verification - Users
- Resources
- Security aspects standards based
16Authentication Solutions for VO Environments
- Single sign on one-time authentication provides
access to allowed Grid resources - Delegation ability to endow a program to
execute on the named users behalf - Interoperate with local security solutions
- User-based trust relationships
17Resources Layer Sharing Single Resources
- Defines protocols, APIs, and SDKs for
- Secure negotiation
- Initiation
- Monitoring
- Control
- Accounting
- Payment processing
- Call Fabric Layer functions to access and control
local resources - Concerned entirely with individual resources
- Primary protocols
- Information protocols obtain information about
state and/or structure of resources - Management protocols negotiate access to a
shared resource
Operations
18Collective Layer Coordinating Multiple Resources
- Protocols, APIs, and SDKs
- Not associated with any one particular resource
- Global in nature
- Capture interactions across collections of
resources
19Collective Layer Services
- Directory services discover existence and/or
properties of VO resources - Co-allocation, scheduling, and brokering services
- Allow VO participants to request allocation of
one or more resources - Allow VO participants to schedule tasks on
appropriate resources - Monitoring and diagnostics services monitor VO
resources for failure, adversarial attack, or
overload - Data replication services support placement of
data to maximize data access performance with
respect to metrics such as response time,
reliability, and cost - Grid-enabled programming services allow use of
familiar programming models to be used in Grid
environments to address resource discover,
security, and resource allocation
20Collective Layer Services
- Workload management systems and collaboration
frameworks - aka problem solving environments
- Provide for description, use, and management of
multi-step, asynchronous, multi-component
workflows - Software discovery services discover and select
most appropriate implementation and execution
platform based on parameters of problem being
solved - Community authorization services enforce
community policies governing resource access, to
generate access capabilities to community
resources - Community accounting and payment services
gather resource usage information for accounting,
payment, and/or resource usage management - Collaboratory services supports information
exchange among users
21Applications
- Utilize services defined at any of the other
layers - Construction
- Utilization
- Implemented using SDKs
- Exchange protocol messages with appropriate
services to perform desired actions - Utilize
- Frameworks
- Libraries
F (well-defined protocols)
22Applications
Key
Collective APIs SDKs
Collective Service Protocols
API/SDK
Collective Services
Resource APIs SDKs
Service
Resource Service Protocols
Resource Services
Connectivity APIs
Connectivity Protocols
Fabric
23Bilateral Relationships
24Multilateral Relationships
Grid
25Grid Architecture Services
26Problem Scenario
- Purchasing CollaborativeAs the result of a
professional society survey initiative, a number
of enterprises in some particular industry
determine that they purchase similar items from a
common set of suppliers. Upon reviewing the
suppliers terms and conditions, a cost saving
benefit is recognized if the several industry
enterprises can engage in a common procurement
activity. From the enterprise perspective, there
would be common pricing based on shared catalogs
price with advantages for larger orders. From
the supplier perspective, there would be fewer
purchase orders to handle, thereby reducing
labor-intensive activities and associated
operating costs. Initial discussions with the
supplier community are encouraging the problem
is that every participant has an individual
information processing environment that offers
minimal commonality. - Could this problem be solved using traditional
computing methods? - What solution possibilities are offered by grid
computing? - Develop a schematic that illustrates a solution
proposal.