Grid Technology and The Latin American Grid

1
Grid Technology (and The Latin American Grid)

• Jaime Seguel
• Department of Electrical and Computer Engineering

2
Outline

• Cyber-infrastructure and Grids
• Overview of Grid Technologies
• Current and future Grid Technology research
• The Latin American Grid

3
Distributed Science and Technology

• Most scientific advancement and breakthroughs
  occur in isolation result from the knowledge,
  imagination and inspiration of a single or a
  small group of scientists. Knowledge and
  expertise are distributed geographically and
  conceptually in specialties.
• Data sources are distributed geographically and
  in specialties
• Computer systems are distributed geographically
  and systems-wise.
• The potential benefits of some degree of
  integration have been voiced in several forums

4
Cyber-infrastructure

• Term coined by NSF to describe ideal research
  environments in which Information Technology
  capabilities are made available to researchers in
  an interoperable computer network.
• Requires orchestrating intelligent searching,
  semantic integration, visualization of
  information, scientific databases and high
  performance computing.
• The expected result is an integrative environment
  for enabling transformative advances in research
  and education.

5
Grid Computing a cyber-infrastructure incarnation

• Grid Computing consists of a very specific set
  of technologies for uniting computing partners
  and systems in a virtual organization.
• Should not be confused with High-Performance,
  Distributed or peer-to-peer Computing.
• Grid computing is intended for inter-organizationa
  l computations.
• Grid Technology is still work in progress.
  Current Grid systems today fall short of Grids
  goals set by the Global Grid Forum (GGF).

6
Grid Technologies

• Grid technology is the software used to create
  Grids, not the network technology.
• A specific set of Grid technologies are the
  outgrowth of the Global Grid Forum (Globus, GSI,
  etc)
• Grids are frequently purpose-built and normally
  built component by component in tight
  collaboration between scientists and IT
  researchers and implementers.
• Turn-key grid systems are not yet available.

7
A Snapshot on the Evolution of Information
Technology

• Internet - Communications backbone based on
  communications protocols (TCP/IP, FTP, HTTP,
  etc.)
• Web Human-Systems interaction for document
  transfer (HTTP) and rendering (HTML, CGI, URI)
• Grid Proposes Human-Systems and Systems-Systems
  interaction for collaborative science, commerce
  and education.

8
World Wide Grid the ultimate cyber-infrastructur
e goal

• WWW is a fine facility for presenting HTML pages
  to people and having them respond but its core
  facilities are not adequate for robust
  computer-to-computer interactions.
• In order to address this deficiency, computer
  scientists began combining concepts from the Web,
  like structured documents and open standards,
  along with traditional distributed computing
  concepts (like RPC and CORBA).
• What follows is a series of steps that complement
  the World Wide Web by providing structured data
  services that are consumed by software rather
  than by humans and that would some day link
  software systems around the globe in a uniform
  manner The World Wide Grid.

9
Current Status (1) XML Protocols

• The foundation for the standards is XML which
  describes data and behavior that can easily move
  between languages, platforms and hardware.
• The XML family set the stage for a primitive
  mechanism to perform distributed systems calls.
• Initial rules for leveraging XML and current
  Internet and Web standards were created (SOAP
  bindings for HTTP, SMTP).

10
Current Status (2) Web Services

• XML was well within reach since it had the same
  basic roots as HTML. However, creating a
  distributed computing infrastructure required a
  special set of protocols.
• A new set of protocols extended XML adding some
  missing computer-to-computer features. Microsoft
  referred them as GXA or Global XML
  Architecture, while IBM called it all Web
  Services.
• This new infrastructure for system-to-system
  communications serves as the foundation for the
  Grid model.

11
Current Status (3) The Service Network

• Like in the Internet, a series of participants
  would have to play a role to make the network
  perform. Routers, gateways, directories, proxies
  they are all needed and not the kind that
  speak TCP/IP they need to understand the new
  web service protocols. This network (sometimes
  called The Service Network) works above the
  Internet, yet relies on it at the most basic
  level.

12
Evolution Summary
13
Just a few examples

• NeuroGrid - analysis of brain activity data
  gathered from the MEG instruments (Japan and
  Australia).
• Particle Physics Data Grid - Integrates
  experiment-specific applications, Grid
  technologies, and facility computation and
  storage resources to form effective end-to-end
  capabilities (ANL, BNL, CalTech, FLNL, Wisonsin,
  UCSD, and others)
• AstroGrid UK scientific grid

14
Current Grid software research

• A meta-data catalogue to find things - GGF
  provides one tailored to the application, takes a
  predicate and returns a logical list of
  needed resources and a Replica Catalogue takes
  that list and returns physical filenames
• Planner - No GGF standard (no automation). Most
  implementations use a Directed Acyclic Graph.
• Scheduler - In GGF, a Scheduler moves files to
  compute hosts and works with an Executor to run
  jobs. Condor is a popular scheduler, there is
  also GRAM (Globus)
• Executor - provide a means to run a job at the
  operating system level. GGF has no standard but
  several are available.
• Data mover - move data between systems GridFTP -
  Argonne National Labs, DataMover - Laurence
  Berkeley Labs, Sabul - U. Illinois and like ftp
  and scp. Low level data movers are often used by
  more sophisticated ones.
• Security - GGF only offers single-sign-on
  Authentication. GGF offers the GSI-GridMap, a
  Kerberos-like, certificate based authentication
  strategy. GGF does not addresses the subject of
  permissions - only authentications!

15
(Near) Future Grid Systems Research

• Coordination and Status A consolidated view of
  the system
• Management - Grid-wide management tools
• Monitoring - Live tracking of processes/objects
• Error recovery and workspace cleanup Policies
  and standards
• Process restartability
• Process management - Grid-wide awareness of
  processes
• Planning (workflow) - Grid-wide awareness and
  automation
• Saving of results - Policy, standards and
  automated transports
• Tracking - Recording of processing, object
  lineage and/or user activity
• Performance metrics
• Data (disk) cache
• Permissions and Object Security

16
Latin American Grid LA Grid
17
LA Grid Mission
Education
Research
Collaboration
Talent Development
18
Why a Latin American Grid?

• The Hispanic minority group is the largest and
  fastest growing ethnic segment in the U.S.
  (currently 14 growing to 25 by 2050).
• Hispanic participation in Computer Science
  Engineering is disproportionately low at the
  Bachelors 3.9 Master 1.3 and PhD 1.0

19
Technology Model (I)

• HW Configuration
• 14 Node Blade Center
• IBM Subsystems
• Dual 32 port SAN switch
• Dual 10 GigE back base switches
• Back end UNIX servers (p/x Series)
• Leverage existing clusters at FIU and UPRM

20
Technology Model (II)

• SW Platform
• ? Grid Aware
• - Middleware
• - Autonomic
• - Fault Tolerance
• - QOS
• Tools
• - Grid Tool kit - DB2
• - Load Leveler - Eclipse
• - Websphere - Linux
• - Rational Suites

21
Organization (I)

• Advisory Board of High Level Executives
• Provide Business, Technology, Diversity
• Direction
• - Chair Pete Martinez
• Nick Bowen
• Irving Wladawsky-Berger
• Ted Childs

22
Organization (II)

• Governance Board
• Provide Operational Guidance
• - Chair Pete Martinez
• Yi Deng, Steve Luis (FIU)
• Jaime Seguel (UPRM)
• Monterrey Tech
• Rosa Badía (Barcelona SC)
• Jean-Pierre Prost (IBM Watson)

23
Current IBM Sponsored Projects Applications Layer

• Bioinformatics on Grid (IBM project leaders
  Howard Ho and Wen-Syan Li)
•  - 2 FTE (Full Time Employee) for IBM researcher
  and
• - 80K, covering 4 summer student internships at
  Almaden Research
• Grid Enablement of Hurricane Mitigation (IBM
  project leader Len Berman)
• - 40K for a full time student from FIU.
• -This work will start in 2006, and it is expected
  to evolve in a PhD thesis

24
Current IBM Sponsored Projects Systems Layer

• Autonomic Resource Management for Heterogeneous
  Grid Environments (IBM project leader Vijay
  Naik)        
• -1 FTE for IBM Research.
• - FIU and UPRM students (4) and faculty (4)
• Meta-scheduling and Job Flow (IBM Project leader
  Liana Fong)
• 1.75 FTE for IBM Research.
• FIU and UPRM students (6) and faculty (6).

25
The Ultimate AimTalent Development

• Executive Mentoring

• Market Value Skills

Highest Potential

• Internships

• Employment

• Research Projects

Top Talent

• Executive Roundtables

• Technology Lectures

• Advanced Education

26

• Thanks!