Virtual Laboratory Overview

1
Virtual Laboratory Overview

• A collaborative analysis environment
• for applied experimental science
• Distributed instrumentation resource access
• Remote resources transparently available
• Focus on content and information
• Guide the user through the experiment

2
Distributed Metacomputing the Grid

• Dependable, consistent and pervasive access to
  (high-end) resources
• Guaranteed end-to-end performance
• Varying resource availability
• Various administrative domains
• Security, policy and payment

3
The Grid, a laymans view

• In ye olde days (till approx 1992)
• Hardly any network security
• All machines in a LAN are created equal
• All local users happy with remote shell, rlogin,
  rcp
• Now
• Want to communicate globally over Gigabit WAN,
  but
• Internet is a dangerous place full of crackers
  and government agencies, firewalls and barriers
• The Grid
• Bring back single sign on and trust
• use the WAN as the 80s LANs all global users
  happy

4
The fabric (machines and network)

• Surfnet5/Gigaport networking
• Now 20 Gbit/s IP backbone (POS framing)
• In 2003 80 Gbit/s
• Connects universities, institutes and acad. home
  users
• Client connections now
• NIKHEF 1 Gbit/s Surfnet,2x1 Gbit/s WTCWnet
  (SARA)
• VU 155 Mbit/s (soon 1Gbit/s)
• Compute Resources
• Farms, supercomputer, tape robot, visualization,

5
Microbeam and VLAM-G network

• Secure connection between microbeam and WTCW
• IPsec tunnel (VPN) encryption and integrity
  checking

SurfNet backbone
VPN router
VPN router
ComputeFarm
CampusNet
Institute Network
VU counting room
6
The VLAM-G Applications

• Three prototype applications for the VLAM-G
• Materials Analysis of Complex Surfaces (MacsLab)
• Microbeam, FT-IR, TOF-SIMS mass spectroscopy
• Biomedical simulation and visualization (VRE)
• Link patient MRI scans with blood-flow
  simulations in vessels
• Genome expression studies using a DNA Micro
  Array (Expressive)
• Mass-test reaction of antibodies on DNA and
  proteins
• Applications share concept of Process Flow
• Many use unique (in NL) resources or need
  compute power

7
A layered architecture
8
Objectives

• Designing middlewarebridge gap between Grid-
  and application-layer
• Enable VL users to define, execute, and monitor
  their experiments
• Provide to VL users
• location independent experimentation,
• familiar experimentation environment
• assistance during his experiment

9
Information and process flow
10
A simple architecture view
KernelDB
11
Application Domain DB

• Characteristics of typical application
• Scientist(s) performing the experiment
• On objects and pre-existing information data
• On which processes operate
• That use apparatus with specific properties
• Resulting in new data and information
• A domain-specific flow of processes

12
VL-AM Kernel DB

• Stores user support information
• experiment topology definitions
• module descriptions
• user information
• Provides cross-links to application
  annotationsknows the context in which data was
  generated
• Extends resource directories now used in Grid

13
Device control

• Needs a very stable and secure environment
• Concurrent access
• Full control for (many) local operators
• Limited control for remote end-users (laymen)
• Control very device dependent (unique properties)
• Use dedicated control software as user interface
• DACQ output needs to be integrated in the VLAM

14
A microprobe experiment (Analysis)
Data storage (tape robot)
Local DACQ system
Local or remote compute power
Local Visualization
15
A Users View
16
VLAM-G current status

• Prototypes of the analysis environment exist
• Distributed compute environment ready _at_WCW (VU
  soon)
• Integrating the various parts of VLAM-G AM
• Currently building the analysis GUI
• Expect working demo in June (HPCN conference)
• Analysis modules Stefan Piet and Gert Eijkel
• Secure Network to VU now selecting equipment
• Building of device interface (LabView) Q3/Q4 2001