The Grid

							The Grid

Background and Architecture
1. Keys to success for IT technologies


            Infrastructure
            Open   Standards
Infrastructure
 without, no one can use the technology
 financed
     by governments, very first
     by industry, after interest increases
Open Standards
 fast evolution
 efficiency will be increased
 more and more integrated
2. How Infrastructure is built
   railroads, telephones and power
   development was very complicated
   started in small regional areas
   connected to a bigger network
   to be successful:
     acceptance  by the users
     support by the governments
          financial
          promotive
3. Scientific demands
 computational vs. observational = 1 to 10
 remote access to
     data
     instrumentation

 data and computation intensive
 powerful management of resources
Problems of experimental science

 only few resources worldwide
 research must be done on site


The Grid may allow
 corporative work between scientists
 team spread all over the world
4. Business Impact
   large corporations are global in extent

The Grid may
 link suppliers, manufactures and
  customers
 unite a company into a single collaborative
  team
Infrastructure for the masses
   only accepted widely if it
     becomes    transparent to the user
     doesn‘t need much knowledge
     is highly reliable
The growth of technology
   development phase
     technology itself is important
     mainly experts

   mass adoption
     applications, reliability and availability
     control returns to experts
          sinks into background
5. History
ARPANET
 started in the early 1970s
 experimental network
 developed important protocols
   TCP/IP
   notion   packet switching
Evolution (1)
   1997, GT2
     usabilityand interoperability
     solutions for authentication and
     resource discovery and access
     protocols, APIs and services
     GT2 „standards“
        not formular
        not for public review
Evolution (2)
   2002, OGSA
     extended  GT2 concepts and technologies
     service-oriented architectures
     Web services
     provides framework
6. Concepts (1)
Analogies to Peer-to-Peer file sharing
 sharing in terms of The Grid
   direct   access to
      computers
      software

      data

      sensors

      all other resources
6. Concepts (2)
  sharing
         under a certain set of rules
  mechanisms for
    accounting
    payment (if needed)
        in money
        in access to user‘s local resources
Concepts (3)
 achieving various QoS
 decomposing of integrated infrastructure
     into   fragmanted systems
   different resources
     shared   under certain circumstances
   pool of resources
     members     can use under a certain set of rules
7. Architecture
   seperated into different layers
     providing    different levels of abstraction
   lowest level
     first   step for resources into the Grid
   core protocol
     establishing    secure connection between Grid
      members
     shared access to local resources
     base for many different applications
Fabric Layer (1)
   provides local resources
     shared  over the Grid
     computational power
     storage
     access to sensors

   translating local protocols to Grid protocols
   components in this layer will act as proxy objects
Fabric Layer (2)
Component
 provides access to one kind of resources
 implements resource specific operations
 general operations for concurrent access
 show higher-level protocols the resources‘
   structure
   state
   capabilities
Connectivity and Resource Layer

   narrow neck (hourglass model)
   based on many maybe different fabric layer
    technologies
   base for many very highspread technologies
   small set of core abstractions and protocols
   local resources connected to those how ask for
    them
Communication protocols
   include
     transport
     routing
     naming

 defined by the ISO/OSI model
 TCP/IP protocol stack
Security (Connectivity Layer)
   one base functionality of this layer
   secure exchange of data
   identity verifying
     users
     resources

   implementations should
     base on existing standards
     support single sign-on
Resource Access (Resource Layer)
   enabling user to interact with remote resources
   defines protocols for
       secure negotiation
       initiation
       monitoring
       control
       accounting
       payment
   information protocols
   managing protocols
Collective Layer
 protocols and services to provide
  interactions across collections of
  resources
 in many cases built inside the application
 for examples
     weather forecast program
     Netsolve/GridSolve 2.0
Application Layer
 comprises the user applications
 applications constructed by calling upon
  services of any layer
 may introduce different layers
      8. Implementations


 Globus Toolkit Version 2 (GT2)
 Open Grid Service Architecture (OGSA)
GT2 (1)
   first standardized implementation
   Grid protocols at higher levels
   assumes suitable software on fabric elements
     CPU    scheduling
     file system management
     sytem monitoring

   some components for discovering information
    about common resource types
GT2 (2)
   connectivity layer defined by GSI protocols
     single  sign-on authentication
     communication protection
     restricted delegation of rights
GT2 (3)
 implements GRAM protocol
 provides secure, reliable creation and
  management of remote computation
 uses
     „gate-keeper“to initiate
     „job manager“ to manage

   „GRAM reporter“ for local computations
GT2 (4)
   Monitoring and Discovery Service (MDS-2)
     discovering   and accessing
        configuration
        status information

     data model
     resource-level protocols
     configurable local registry
     collective registry
OGSA (1)
 standardization of core GT protocols
 use essential Grid functions in different
  settings
 service orientated
 uniform treatment of all network entries
OGSA (2)
   Grid service
     implements
        standard interfaces
        behaviors

        conventions

     services   are defined by the OGSI
Resources
   Need some sort of refundment
     Financial
     other
P2P-Networks: eMule
 Refundment by Priority
 Modifier * Waitingtime => Queue Rank
Credit System in eMule
 Ratio1 = 2*Up / Down
 Ratio2 = SQRT(Up+2)
 Modifier = Min{Ratio1,Ratio2}
 1<=Modifier<=10
 If Up < 1 MB => Modifier = 1
 If Down = 0 => Modifier = 10
OurGrid
 CPU-Sharing
 Round based
 Problems:
     Free Riders
     ID Changers
rA(B) = v(B,A)−v(A,B)
 rA(B): reputation of B relative to A
 v(B,A): Value of favours B done to    A
 v(A,B): Value of favours A done to B
 rho: probability of consumer in turn
 f: probability of freerider
 epsilon: probability of a freerider getting a
  resource
rA(B) = v(B,A)−v(A,B)




                        f=0,5
rA(B) = v(B,A)−v(A,B)




                        rho=0,5
rA(B) = v(B,A)−v(A,B)




                        rho=0,5
rA(B) = max{0, v(B,A) - v(A,B)}




                             rho=0,5
rA(B) = max{0, v(B,A) - v(A,B) + log(v(B,A))}




                                        rho=0,5
Thank You