Computing as a Peer to Peer Grid Service

Computing as a Peer to Peer Grid Service PTLIU Laboratory for Community Grids Geoffrey Fox Computer Science, Informatics, Physics Indiana University Bloomington IN 47404 gcf@indiana. edu 9/26/2020 gecdistedaug 01 1

Some Technology Trends l l l l Increasing performance of Internet backbone and last mile (access) Hand-held devices and wireless Pervasive Access Peer to peer technologies enable new ways of collaborating and blurs distinction between clients and servers Client-Server Multi-tier Architectures XML Schema and tools All data defined as objects Separation of client, system and persistent storage models for information Development of (application) service model to capture common (maybe centralized) capabilities Semantic Web, Grid or … “Next Generation Web” 9/26/2020 gecdistedaug 01 2

Small Devices Increasing in Importance l l l There is growing interest in wireless portable displays in the confluence of cell phone and personal digital assistant markets By 2005, 60 million internet ready cell phones sold each year 65% of all Broadband Internet accesses via non desktop appliances 9/26/2020 CM 5 gecdistedaug 01 3

Technology Trends and Principles l l l All performance and capability measures of infrastructure continue to improve Gilder’s law says that network bandwidth increases 3 times faster than CPU Performance (Moore’s Law) The Telecosm eclipses the Microcosm …. George Gilder Telecosm : How Infinite Bandwidth Will Revolutionize Our World (September 2000, Free Press; ISBN: 0684809303, #146(3883) in Amazon Sales Jan 15 2001(July 29 2001)) 9/26/2020 gecdistedaug 01 4

What is a Grid Service? l l The Grid is distributed system allowing communities to access seamlessly heterogeneous resources from heterogeneous clients – Resources are web-pages, instruments, Object repositories, Simulation codes running on supercomputers …. A Service is a generic application or capability respecting standards (general web and application specific) allowing multiple providers to compete on a given service Middle Tier Portal is Back end Broker customizable Capabability User interface The Grid is essentially is the future Web Resource IBM just announced they were investing around $1 Billion in Grid 9/26/2020 gecdistedaug 01 5

Some General Grid Services l Business is developing “web service” concept to support areas like e-commerce where one composes atomic services like – – Security Payment Catalog Goods supply Each of these services could allow Multiple choices of provider In a given session Payment Credit Card Security WSDL is new standard for web services 9/26/2020 gecdistedaug 01 Catalog Warehouse shipping 6

Grid Services support Communities l l l Grid Communities (PTLIU, NSF, Earth Science, High School Classes) are groups of communicating individuals sharing resources implemented as Grid Services Access Grid from Argonne/NCSA is best Audio/Video conferencing technology Peer to Peer networking describes a set of technologies supporting community building with an emphasis on less structured groups than classic “users of a supercomputer” Peer to peer Grids combine the technologies and support “small worlds” – optimized networks with short links between each community member Collaborative Grid Service Framework allows one to build community not individually oriented Grid Services 9/26/2020 gecdistedaug 01 7

Architecture of Grid: Commodity Science l Commerce, Entertainment, Healthcare, Science, Computing, Education …. will be Grid Services C o n v e n i e n c e Next Generation Consumer Web Twenty-First Century University and laboratory Community Portals Science Portals & Workbenches Commerce Grid Education Grid Research Grid Computational Grid Business Services Education Services Research Services & Technology Computational Services Grid Services (resource independent) Grid Fabric (resource dependent) P e r f o r m a n c e Networking, Devices and Systems 9/26/2020 gecdistedaug 01 8

Examples of Grid or Web Services l l There are generic Grid system services: security, collaboration, persistent storage, universal access An Application Service is a capability used either by another service or by a user – It has input and output ports – data is from sensors or other services Consider NASA Space Operations (CSOC) as a Grid Service – Spacecraft management (with a web front end) – Each tracking station is a service – Image Processing is a pipeline of filters – which can be grouped into different services – Data storage is an important system service – Big services built hierarchically from “basic” services Portals are the user (web browser) interfaces to Grid services 9/26/2020 gecdistedaug 01 9

Integration of Grid Services Multidisciplinary Control Parallel DB Proxy Database Image Processing Server Grid Gateway Supporting Seamless Interface Sensor Control Data Mining Server Origin 2000 Proxy Net. Solve Linear Alg. Server Agent-based Choice of Compute Engine Matrix Solver IBM SP Proxy Object Grid Programming Environment 9/26/2020 MPP Classic HPCC Resources 10 gecdistedaug 01

The Application Service Model l l As bandwidth of communication (between) services increases one can support smaller services Some fields such as Education do not have stringent latency/bandwidth requirements on inter-service communication – Computing services must often have high performance communication l l A service “is a component” and is a replacement for a library in case where performance allows Services are a sustainable model of software development – each service has documented capability with standards compliant interfaces – XML defines interfaces at several levels – WSDL at Grid level and XSIL or equivalent for scientific data format l l A service can be written in Perl, Python, Java Servlet, Enterprise Javabean, CORBA (C++ or Fortran) Object … Communication protocol can be RMI (Java), IIOP (CORBA) or SOAP (HTTP, XML) …… 9/26/2020 gecdistedaug 01 11

Classic Grid Architecture Resources Database Composition Neos Netsolve Security Portal Middle Tier Brokers Service Providers Portal Typically separate Clients Servers Resources 9/26/2020 Clients gecdistedaug 01 Users and Devices 12

Peers Peer to Peer Network User Service Resource Routing Peers are Jacks of all Trades linked to “all” peers in community Typically Integrated Clients Servers and Resources User Service Resource Routing 9/26/2020 User Service Resource Routing gecdistedaug 01 User Service Resource Routing 13

Peer to Peer Grid User Service Resource Routing GMS Routing User Service Resource Routing 9/26/2020 Services Dynamic Message or Event Routing from Peers or Servers User Service Resource Routing gecdistedaug 01 14

P 2 P Grid with Peers and Servers at the center of the world Peers on the Edge of the Internet 9/26/2020 gecdistedaug 01 15

HPCC Background l l The 1990 HPCC 10 year initiative was largely aimed at enabling large scale simulations for a broad range of computational science and engineering problems It was in many ways a success and we have methods and machines that can (begin to) tackle most 3 D simulations – ASCI simulations particularly impressive – Do. E still putting substantial resources into basic software and algorithms from adaptive meshes to PDE solver libraries l l Machines are still increasing in performance exponentially and should achieve petaflops in next 7 -10 years Each computing community needs to harness these capabilities in customized fashion – ASCI(Do. E), Earth Simulator(Japan), Teragrid(NSF) …. . 9/26/2020 gecdistedaug 01 16

Some HPCC Difficulties l l An Intellectual failure: we never produced a better programming model than message passing – HPCC code is hard work – “High point” of ASCI software is “Grid FTP” An institutional problem: we do not have a way to produce complex sustainable software for a niche (1%) market like HPCC. – POOMA support just disappeared one day – Do. E is funding efforts for their critical missions – not to support general communities – One must adopt commodity standards and produce “small” sustainable modules. – Note distributed memory becoming dominant again with complex hybrid clustered SMP architecture – not clear that “wise” to exploit advantages of shared memory architectures 9/26/2020 gecdistedaug 01 17

Personal HPCC Advice l l l l KISS: Keep it Simple and Sustainable Use MPI and open. MP if needed for performance on shared memory nodes Adaptive Meshes Are well understood Load Balancing to get high performance parallel simulations PDE Solvers including fast multipoles Use broad community expertise Particle dynamics Other areas such as datamining, visualization and data assimilation quite advanced but still significant research 9/26/2020 } gecdistedaug 01 18

Use of Object Technologies l l The claimed commercial success in using Object and component technology has not been a clear success in HPCC – Object technologies do not naturally support either high performance or parallelism – C++ can be high performance but CORBA and Java are not – There is no agreed HPCC component architecture to produce more modern libraries (Do. E has very large CCA – Common Component Architecture – effort which should be followed) Fortran will continue to decline in importance and interest – the community should prefer not to use it – It’s use will not attract the best students 9/26/2020 gecdistedaug 01 19

Application Structure l Modern applications are typically multi-scale and multi -disciplinary – i. e. a given simulation is made of multiple components with either different time/length scales and/or multiple authors from possibly multiple fields l l I am not aware of a systematic “Computational renormalization group” – a methodology that links different scales together However composition of modules is an area where technology of growing sophistication is becoming available – Needed commercially to integrate corporate functions – CCA tackles challenging “small grain size”; Gateway example of clearly successful large grain size integration 9/26/2020 gecdistedaug 01 20

Object Size & Distributed/Parallel Simulations l All interesting systems consist of linked entities – Particles, grid points, people or groups thereof l Linkage translates into message passing – – – l l l Cars on a freeway Phone calls Forces between particles Amount of communication tends to be proportional to surface area of entity whereas simulation time proportional to volume So communication/computation is surface/volume and decreases in importance as entity size increases In parallel computing, communication synchronized; in distributed computing “self contained objects” (whole programs) which can be scheduled asynchronously 9/26/2020 gecdistedaug 01 21

Community HPCC and Grid Strategy I l Decide what services are well enough understood and useful enough to be encapsulated as application services – – l l Parallel FEM Solvers Visualization Parallel Particle Dynamics Access to Sensor Data Make as small as possible – smaller is simpler and more sustainable but with higher communication needs Establish teams to design and build services Use a framework offering needed Grid System services Build electronic community for each field with collaboration tools, resources and world wide networking linking community members 9/26/2020 gecdistedaug 01 22

Community HPCC and Grid Strategy II l l l Some capabilities – such as fast multipole or adaptive grids package – should be built as classic libraries or templates Other services – such as datamining or support of multi -scale simulations – need research using a toolkit approach if one can design a general structure Need “hosts” for major services – access and storage of sensor data Need funds to build and sustain “infrastructure” and research services Use electronic community tools to enhance Collaboration 9/26/2020 gecdistedaug 01 23

Sensor Grid Service out port universal sensor access people/computers 9/26/2020 Distributed Sensor Service in ports gecdistedaug 01 24

Peer to Peer Grid Community APAN Network linking gecdistedaug 01 Access Grids 9/26/2020 25