Research Achievements Kenji Kaneda Agenda Research background and
Research Achievements Kenji Kaneda
Agenda Research background and goal n Overview of my research achievements n Phoenix n Virtual Private Grid n n Summary and recent activities
Research Background and Goal
Background n Grid computing Parallel computing with harnessing many widely-distributed resources E. g. ) aggregate of PC clusters spread over multiple LANs n
Traditional Parallel Computing vs. Grid Computing Traditional parallel computing Reliable processors n Single-LAN resources n Grid computing Unreliable processors n Multi-LANs resources n
Difficulty in Grid computing n Frequent machine/network failures E. g. ) 1 machine failure per a day n Restricted Connectivity Administrative policies restrict communications between machines E. g. ) firewall, NAT, DHCP n Firewall TCP Gateway
Research Goal n Allow a user to harness a computational grid like traditional parallel computing Fault tolerance n Transparent communication on WANs n
My Research Achievements Design/implementation of middlewares n Phoenix n n Parallel programming library for accommodating dynamically joining/leaving resources Virtual Private Grid n Command shell for utilizing hundreds of computers spread over multiple LANs
Phoenix
Phoenix n Parallel programming library for accommodating dynamically joining/leaving resources Programming model for supporting migration of application states n Transparent communication mechanism for WANs n
Programming Model for Supporting Migration of Application States Subsumes a regular message passing model n Provides a namespace that does not depend on physical machines n n n Programmer uses this name to specify a message destination Programmer can write a program without being aware of physical machines
Transparent Communication mechanism for WANs Overlay network construction n Application-level message routing n n Processes can communicate with one another even if networks are not fully connected n even if connection topologies change dynamically n
Demonstration Boot processes on 3 subnets n Add processes dynamically n
Demonstration
Experiments (1/3) n Speedup with fixed resources POV-Ray: 78 speedup using 104 processors on 3 LANs n LU: comparable to MPICH (on a single LAN) n
Experiments (2/3) n Speedup with dynamic resources n POV-Ray takes advantage of dynamically added resources quickly
Experiments (3/3) n Parallel shogi (Japanese chess) program on 720 laptop PCs n 7~8 speedup
Related Work n Grid enabled MPIs E. g. ) MPICH-G [G. Bosilca et al. SC’ 02] n Based on a traditional message passing model n Difficult to support dynamic changes of resources n Communications libraries for Grids E. g. ) Ibis [A. Denis et al. HPDC’ 04] n Static message routing
Summary ~ Phoenix ~ n Parallel programming library for dynamically changing resources n Good speedup with a large number of machines on multiple LANs
Virtual Private Grid
Virtual Private Grid (VPG) n Command shell for utilizing hundreds of computers spread over multiple LANs
Features (1/2) n User can submit jobs without caring administrative restrictions E. g. ) cmd 1@host 1 | cmd 2@host 2 > file 3@host 3 Write to file 3 Firewall NAT host 3 Firewall Execute cmd 1 host 1 Execute cmd 2 host 2
Features (2/2) n Fault tolerance VPG can continue to run even if some machines are added/deleted dynamically n No central server is required n
Demonstration n Environment 3 LANs n CPU: Sparc, x 86, MIPS, Power. PC n OS: Solaris, Linux, IRIX n
Demonstration
Related Work n Grid job submission tools E. g. ) Globus, Condor-G n Difficult to submit jobs to machines under administrative restrictions
Summary ~ Virtual Private Grid ~ n Command shell for utilizing hundreds of computers spread over multiple LANs n Fast job submission to more than 100 machines
Summary and Recent activities
Summary ~ My Research Achievements ~ n Middlewares for Grid computing Phoenix n Virtual Private Grid n
Recent Activities (1/2) n Virtual SMP n Emulates a multi-processor machines on a loosely-coupled computers Virtual dual processor machine on two single processor machines
Recent Activities (2/2) n Virtual SMP n Easy utilization of distributed resources with a common OS (e. g. , Windows, Linux)
- Slides: 31