The OptIPuter Project – Removing Bandwidth as an Obstacle In Data Intensive Sciences

26
The OptIPuter Project – Removing Bandwidth as an Obstacle In Data Intensive Sciences Opening Remarks OptIPuter Team Meeting University of California, San Diego February 6, 2003 Dr. Larry Smarr Director, California Institute for Telecommunications and Information Technologies Harry E. Gruber Professor, Dept. of Computer Science and Engineering Jacobs School of Engineering, UCSD

description

The OptIPuter Project – Removing Bandwidth as an Obstacle In Data Intensive Sciences. Opening Remarks OptIPuter Team Meeting University of California, San Diego February 6, 2003. Dr. Larry Smarr Director, California Institute for Telecommunications and Information Technologies - PowerPoint PPT Presentation

Transcript of The OptIPuter Project – Removing Bandwidth as an Obstacle In Data Intensive Sciences

Page 1: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

The OptIPuter Project – Removing Bandwidth as an Obstacle

In Data Intensive Sciences

Opening Remarks

OptIPuter Team Meeting

University of California, San Diego

February 6, 2003

Dr. Larry Smarr

Director, California Institute for Telecommunications and Information Technologies

Harry E. Gruber Professor,

Dept. of Computer Science and Engineering

Jacobs School of Engineering, UCSD

Page 2: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

The Move to Data-Intensive Science & Engineering-e-Science Community Resources

ATLAS

Sloan Digital Sky Survey

LHC

ALMA

Page 3: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Why Optical Networks Are Emerging as the 21st Century Driver for the Grid

Scientific American, January 2001

Parallel Lambdas Will Drive This DecadeThe Way Parallel Processors Drove the 1990s

Page 4: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

CONTROL

PLANE

Clusters

DynamicallyAllocatedLightpaths

Switch Fabrics

PhysicalMonitoring

Apps Middleware

A LambdaGrid Will Be the Backbone for an e-Science Network

Source: Joe Mambretti, NU

Page 5: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

The Biomedical Informatics Research Network a Multi-Scale Brain Imaging Federated Repository

BIRN Test-bedsBIRN Test-beds::Multiscale Mouse Models of Disease, Human Brain Morphometrics, and Multiscale Mouse Models of Disease, Human Brain Morphometrics, and

FIRST BIRN (FIRST BIRN (10 site project for fMRI’s of Schizophrenics)10 site project for fMRI’s of Schizophrenics)

NIH Plans to Expand to Other Organs

and Many Laboratories

Page 6: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

GEON’s Data Grid Team Has Strong Overlap with BIRN and OptIPuter

• Learning From The BIRN Project– The GEON Grid:

– Heterogeneous Networks, Compute Nodes, Storage

– Deploy Grid And Cluster Software Across GEON– Peer-to-Peer Information Fabric for Sharing:

– Data, Tools, And Compute Resources

Source: Chaitan Baru, SDSC, Cal-(IT)2

Two Science “Testbeds” Broad Range Of Geoscience Data Sets

NSF ITR Grant $11.25M

2002-2007

Page 7: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

NSF’s EarthScopeRollout Over 14 Years Starting

With Existing Broadband Stations

Page 8: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Data Intensive Scientific Applications Require Experimental Optical Networks

• Large Data Challenges in Neuro and Earth Sciences– Each Data Object is 3D and Gigabytes– Data are Generated and Stored in Distributed Archives– Research is Carried Out on Federated Repository

• Requirements– Computing Requirements PC Clusters– Communications Dedicated Lambdas Over Fiber– Data Large Peer-to-Peer Lambda Attached Storage – Visualization Collaborative Volume Algorithms

• Response– OptIPuter Research Project

Page 9: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Coherence

DRAM - 4 GB - HIGHLY INTERLEAVEDMULTI-LAMBDAOptical Network

VLIW/RISC CORE40 GFLOPS

10 GHz

240 GB/s24 Bytes wide

240 GB/s24 Bytes wide

VLIW/RISC CORE 40 GFLOPS 10 GHz

...

2nd LEVEL CACHE8 MB

2nd LEVEL CACHE 8 MB

CROSS BAR

DRAM – 16 GB64/256 MB - HIGHLY INTERLEAVED

640GB/s

OptIPuter Inspiration--Node of a 2009 PetaFLOPS Supercomputer

Updated From Steve Wallach, Supercomputing 2000 Keynote

5 Terabits/s

Page 10: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Global Architecture of a 2009 COTS PetaFLOPS System

I/O

ALL-OPTICAL SWITCH

Multi-DieMulti-Processor

1

23

64

63

49

48

4 516

17

18

32

3347 46

128 Die/Box4 CPU/Die

10 meters= 50 nanosec Delay

...

...

...

...

LAN/WAN

Source: Steve Wallach, Supercomputing 2000 Keynote

Systems Become GRID Enabled

Page 11: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

From SuperComputers to SuperNetworks--Changing the Grid Design Point

• The TeraGrid is Optimized for Computing– 1024 IA-64 Nodes Linux Cluster– Assume 1 GigE per Node = 1 Terabit/s I/O– Grid Optical Connection 4x10Gig Lambdas = 40 Gigabit/s– Optical Connections are Only 4% Bisection Bandwidth

• The OptIPuter is Optimized for Bandwidth– 32 IA-64 Node Linux Cluster– Assume 1 GigE per Processor = 32 gigabit/s I/O– Grid Optical Connection 4x10GigE = 40 Gigabit/s– Optical Connections are Over 100% Bisection Bandwidth

Page 12: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Convergence of Networking Fabrics

• Today's Computer Room– Router For External Communications (WAN)– Ethernet Switch For Internal Networking (LAN)– Fibre Channel For Internal Networked Storage (SAN)

• Tomorrow's Grid Room– A Unified Architecture Of LAN/WAN/SAN Switching– More Cost Effective

– One Network Element vs. Many

– One Sphere of Scalability– ALL Resources are GRID Enabled

– Layer 3 Switching and Addressing Throughout

Source: Steve Wallach, Chiaro Networks

Page 13: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

½ Mile

The UCSD OptIPuter Deployment

SIO

SDSC

CRCA

Phys. Sci -Keck

SOM

JSOE Preuss

6th College

Phase I, Fall 02

Phase II, 2003

SDSCAnnex

Collocation point

Node M

The OptIPuter Experimental UCSD Campus Optical Network

Earth Sciences

SDSC

Arts

Chemistry

Medicine

Engineering

High School

UndergradCollege

Phase I, Fall 02

Phase II, 2003

SDSCAnnex

To CENIC

Collocation point

Collocation

Chiaro Router

Production Router

Source: Phil Papadopoulos, SDSC; Greg Hidley, Cal-(IT)2

Page 14: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Metro Optically Linked Visualization Wallswith Industrial Partners Set Stage for Federal Grant

• Driven by SensorNets Data– Real Time Seismic– Environmental Monitoring – Distributed Collaboration– Emergency Response

• Linked UCSD and SDSU– Dedication March 4, 2002

Linking Control Rooms

Cox, Panoram,SAIC, SGI, IBM,

TeraBurst NetworksSD Telecom Council

UCSD SDSU44 Miles of Cox Fiber

Page 15: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

National Light Rail- Serving Very High-End Experimental and Research Applications

• Extension of CalREN-XD Dark Fiber Network– Serves Network Researchers in California Research

Institutions– Four UC Institutes, USC/ISI, Stanford and CalTech

– 10Gb Wavelengths (OC-192c or 10G LANPHY) – Dark Fiber– Point-Point, Point-MultiPoint 1G Ethernet Possible

• NLR is a Dark Fiber National Footprint– 4 - 10GB Wavelengths Initially– Capable of 40 10Gb Wavelengths at Build-Out– Partnership model

John Silvester, Dave Reese, Tom West-CENIC

Page 16: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

National Light Rail Footprint Layer 1 Topology

PITPIT

PORPOR

FREFRE

RALRAL

WALWAL

NASNASPHOPHO

STHSTHATLATL

CHICHI

CLECLE

KANKAN

OGDOGD

SACSAC BOSBOSNYCNYC

WDCWDC

STRSTR

DALDAL

DENDEN

LAXLAX

SVLSVL

SEASEA

SDGSDG

JACJAC

15808 Terminal, Regen or OADM site (OpAmp sites not shown)

Fiber route

John Silvester, Dave Reese, Tom West-CENIC

Page 17: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Calient Lambda Switches Now Installed at StarLight and NetherLight

GigE = Gigabit Ethernet (Gbps connection type)

8-processor cluster

16-processor cluster

Switch/Router

8 GigE16 GigE

8 GigE16 GigE

Control plane

Data plane

“Groomer” at StarLight

8 GigE

2 GigE

128x128MEMS

Optical Switch

N E T H E R L I G H T

16-processor cluster

8 GigE

16 GigE

16 GigE

“Groomer” at NetherLight

Control plane

Data plane

2 GigE

OC-192

(10Gbps)

64x64MEMS

Optical Switch

Switch/Router

GigE = Gigabit Ethernet (Gbps connection type)

8-processor cluster

16-processor cluster

Switch/Router

8 GigE16 GigE

8 GigE16 GigE

Control plane

Data plane

“Groomer” at StarLight

8 GigE

2 GigE

128x128MEMS

Optical Switch

N E T H E R L I G H T

16-processor cluster

8 GigE

16 GigE

16 GigE

“Groomer” at NetherLight

Control plane

Data plane

2 GigE

OC-192

(10Gbps)

64x64MEMS

Optical Switch

Switch/Router

Source: Maxine Brown

Page 18: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Amplified Collaboration Environments

Collaborative Tiled Display Accessgrid Multisite

Video Conferencing

CollaborativePassive Stereo

Display

CollaborativeTouch Screen

Whiteboard

WirelessLaptops &

Tablet PCs To Steer The Displays

Source: Jason Leigh

Page 19: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

The OptIPuter 2003

Experimental NetworkWide Array of Vendors

Page 20: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

OptIPuter Software Research

• Near-term Goals: – Build Software To Support Applications With Traditional Models

– High Speed IP Protocol Variations (RBUDP, SABUL, …)– Switch Control Software For DWDM Management And Dynamic Setup– Distributed Configuration Management For OptIPuter Systems

• Long-Term Goals: – System Model Which Supports:

– Grid– Single System– Multi-System Views

– Architectures Which Can: – Harness High Speed DWDM– Exploit Flexible Dispersion Of Data And Computation

– New Communication Abstractions & Data Services – Make Lambda-Based Communication Easily Usable– Use DWDM to Enable Uniform Performance View Of Storage

Source: Andrew Chien, UCSD

Page 21: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Photonic Data Services & OptIPuter

1. Physical

4. Transport – TCP, UDP, SABUL,… (USC,UIC)

5b. Data Services – SOAP, DWTP, (UIC/LAC)

6. Data Intensive Applications (UCI)

2. Photonic Path Serv. – ODIN, THOR,... (NW)

3. IP

5a. Storage (UCSD)

Source: Robert Grossman, UIC/LAC

Page 22: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

OptIPuter is Exploring Quanta as a High Performance Middleware

• Quanta Is A High Performance Networking Toolkit / API

• Quanta Uses Reliable Blast UDP:– Assumes An Over-Provisioned Or Dedicated Network– Excellent For Photonic Networks – Don’t Try This On Commodity Internet!

– It Is Fast!

– It Is Very Predictable

– We Give You A Prediction Equation To Predict Performance

– It Is Most Suited For Transferring Very Large Payloads

• RBUDP, SABUL, and Tsunami Are All Similar Protocols That Use UDP For Bulk Data Transfer

Source: Jason Leigh, UIC

Page 23: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

XCP Is A New Congestion Control SchemeWhich is Good for Gigabit Flows

• Better Than TCP – Almost Never Drops Packets– Converges To Available Bandwidth Very Quickly, ~1Round Trip Time– Fair Over Large Variations In Flow Bandwidth and RTT

• Supports existing TCP semantics– Replaces Only Congestion Control, Reliability Unchanged– No Change To Application/Network API

• Status– To Date: Simulations and SIGCOMM Paper (MIT).

– See Dina Katabi, Mark Handley, and Charles Rohrs, "Internet Congestion Control for Future High Bandwidth-Delay Product Environments." ACM SIGCOMM 2002, August 2002. http://ana.lcs.mit.edu/dina/XCP/

– Current: – Developing Protocol, Implementation – Extending Simulations (ISI)

Source: Aaron Falk, Joe Bannister, ISI USC

Page 24: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Multi-Lambda Security Research

• Security Frequently Defined Through Three Measures: – Integrity, Confidentiality, And Reliability (”Uptime”)

• Can These Measures Can Be Enhanced By Routing Transmissions Over Multiple Lambdas Of Light?

• Can Confidentiality Be Improved By Dividing The Transmission Over Multiple Lambdas And Using “Cheap” Encryption?

• Can Integrity Be Ensured Or Reliability Be Improved Through Sending Redundant Transmissions And Comparing?

Source: Goodrich, Karin

Page 25: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Research on Developing an Integrated Control Plane

OpticalLambda Switching

LogicalLabel

Switching

OpticalBurst

Switching

Integrated Control Plane

Megabit Stream

Gigabit Stream

Bursty Traffic

Multiple User Data Planes

Lambda Inverse

Multiplexing

Tera/Peta Stream

Source: Oliver Yu, UIC

Page 26: The OptIPuter Project –  Removing Bandwidth as an Obstacle  In Data Intensive Sciences

Fast polygon and volume rendering with stereographics

GeoWall

Earth Science

GeoFusion GeoMatrix Toolkit

Underground Earth Science

Rob Mellors and Eric Frost, SDSUSDSC Volume Explorer

Dave Nadeau, SDSC, BIRNSDSC Volume Explorer

NeuroscienceAnatomy

Visible Human ProjectNLM, Brooks AFB,

SDSC Volume Explorer

3D APPLICATIONS:

+

=

OptIPuter Transforms Individual Laboratory Visualization, Computation, & Analysis Facilities

The Preuss School UCSD OptIPuter Facility