The MAGIC Project Final Demonstration and Briefing

1
The MAGIC ProjectFinal Demonstration and Briefing

• Ira Richer
• 29 July 1999

2
MAGIC Scenario
3
Datasets

• Globe
• San Francisco Bay area
• Menlo Park
• Ft. Irwin
• Camp Pendleton (low resolution)
• Camp Pendleton (high resolution)
• Laguna Beach
• Yosemite
• Colorado
• Lawrence, KS
• Ohio
• Washington, DC
• Ft. Benning

4
Background

• The MAGIC scenario is based on applications such
  as mission planning and crisis management
• These applications need to
• locate, process, and display real-time and/or
  archived datasets on-the-fly
• fuse multiple types of data (e.g., weather,
  terrain)
• be able to transfer large volumes of data for
  interactive visualization
• support a range of user capabilities (e.g.,
  command center, mobile user)
• MAGIC focuses on a terrain visualization
  application that uses geo-referenced datasets

5
Objectives

• To develop an information system that provides
  application users with fast, interactive access
  to very large volumes of remote data distributed
  over a wide geographic area
• To demonstrate the above capability in the
  context of a 3-D terrain visualization application

6
MAGIC Participants

• Corp. for National Research Initiatives
• USGS/EROS Data Center
• Lawrence Berkeley National Laboratory
• Sprint
• SRI International
• University of Kansas

7
Distinguishing Characteristics

• Distributed information system, designed from the
  outset to work over a network
• adaptation to network conditions and to system
  load
• No limit on size of viewable datasets
• No restrictions on location of datasets
• Fusion of multiple datasets
• Near real-time visualization of terrain data

8
MAGIC Components

• High-speed internetwork
• Distributed parallel storage system (DPSS)
• Software for terrain dataset preparation
• 3-D interactive terrain visualization
  application TerraVision

9
Generic Configuration
High-speed internetwork
10
System Operation
Server 1
Internetwork
11
Todays Demonstration and Briefing

• 3-D, interactive visualization of large volumes
  of remote data
• Platform scalability
• On-demand data preparation and 3-D flyover
• Fusion of multiple types of data from multiple
  sources
• DPSS fault tolerance

12
Todays Configuration
ESnet
Abilene
GPN
KU
LBNL
EDC
NREN
DARPA
ISI-E
ATDnet
NASA Goddard
13
Dataset Characteristics
14
Data Flow (1)
Abilene
ESnet
GPN
KU
LBNL
EDC
NREN
DARPA
ISI-E
ATDnet
NASA Goddard
15
Dataset Characteristics
16
Platform Scalability

• SGI Octane (50k)
• OC-3, IRIX, DPSS
• 60 frame/sec, 80Mb/s
• Dual-processor PC with graphics accel. (8k)
• Fast Ethernet, Windows NT, HTTP
• 30 frame/sec, 0.5Mb/s (Internet-limited)
• Dual-processor PC (4k)
• OC-3, Linux, DPSS
• 10 frame/sec, 10Mb/s
• Laptop (3k)
• Fast Ethernet, Windows 98, HTTP
• 4 frame/sec, 0.5Mb/s (Internet-limited)

17
TerraVision II

• Composite datasets data from multiple sources,
  at different resolutions
• VRML overlays wind vectors, buildings, roads
  viewable with standard browser
• Wind-vector dataset hosted at NAVO
• Co-registered maps for navigation
• Pre-defined flight paths for operations support,
  testing

18
DPSS Cache Architecture
19
DPSS Costs (Jan. 99)
20
Monitoring and measurement

• Real-time traffic monitor
• DPSS system monitor
• NetLogger
• Logs events at critical points of distributed
  applications
• Provides complete view of system operation
• Visualization tools allow for real-time
  monitoring
• Agent-based management of measurement tools

21
NetLogger Instrumentation
22
DPSS Capabilities

• Load balancing
• Dataset must be replicated
• Tile requests sent to server that will provide
  the best performance, based on current server
  load and network conditions
• Automatic TCP tuning
• Window size determined by throughput and delay
  information
• Fault tolerance
• Automatic detection of failed server, switchover
  to alternate servers, and reconnection after
  recovery

23
Data Flow (2)
ESnet
Abilene
GPN
KU
LBNL
EDC
NREN
DARPA
ISI-E
ATDnet
NASA Goddard
24
Technology Transfer and Spinoffs

• DPSS
• Earth Systems Grid (climate modeling)
• Kaiser Permanente project
• Combustion Corridor (diesel modeling)
• Matisse (MEMS/Supernet)
• NetLogger
• Being used by many network and grid projects

25
Technology Transfer and Spinoffs (concluded)

• USGS archive
• Web-based access to digital earth-science data
• Gateway to the Earth Ohio Pilot (automatic
  dissemination of Landsat imagery)
• TerraVision
• BADD (Dissemination of VRML terrain datasets to
  warfighter)
• Digital Earth (fast indexing of geo-data,
  multimodal capabilities, and education
  applications)
• Netscape plug-in