Doing an MSc Thesis in the Group Parallel and Distributed Systems D'H'J' Epema Parallel and Distribu

1
Doing an MSc Thesis in the GroupParallel and
Distributed SystemsD.H.J. Epema Parallel and
Distributed Systems Group (PDS)Department of
Software TechnologyFaculty EEMCS

2
Contents

• The PDS group
• Research topics in the PDS group
• Procedure for doing an MSc in the PDS group

3
Parallel and Distributed Systems (1)

• Faculty
• prof.dr.ir. H.J. Sips (full professor, chair)
• ir.dr. D.H.J. Epema (associate professor)
• dr. K.L. Langendoen (associate professor)
• dr.ir. J.A. Pouwelse (assistant professor)

4
Parallel and Distributed Systems (2)

• PhD students and PostDocs
• 10 PhD students, 4 PostDoc, 4 programmers
• MSc students
• 12 MSc students (Computer Science, Computer
  Engineering, Embedded Systems)
• Location 9th floor

5
What is PDS about?

• We make resources work
• CPUs, memories, power, networks
• resource management, performance analysis
• Internet applications peer-to-peer systems
  (P2P)
• Middleware grids
• Networked systems wireless sensor networks
  (WSN)
• HPC systems parallel systems

Common theme of our research new concepts gt
design gt implementation gt analysis
6
Research topic 1 Peer-to-peer systems

• In most DSs, some nodes have more authority or
  functionality than others (e.g., in a
  client-server system)
• In peer-to-peer systems, all nodes have the same
  authority and functionality, i.e., these systems
  are decentralized
• Most P2P systems used for file sharing
• Topics in P2P
• Searching for content
• Downloading
• Dynamics
• Scalability
• Freeriding

P2P systems account for more than 50 of the
Internet traffic
7
P2P TV distribution over the Internet

• Distributing TV is the killer P2P application in
  the internet in the next decade
• recorded millions of PVRs form one huge
  repository
• (how to find things)
• live low-cost entry for content distributors
• (how to stream things)
• P2P-TV forms a foundation for sharing with your
  friends (creating virtual communities)
• content (you can have what I have)
• interest profiles (you may like what I like)
• P2P-TV is a viable and innovation-driving
  alternative to (server-client) IP-TV

8
P2P Research in Delft

• Tribler
• Is based on the Bittorrent P2P file-sharing
  system
• Looks at the peers as really representing actual
  users rather than as anonymous computer programs
• Adds social-based functionality
• De-anonymizes peers
• peers have a quasi-unique public permanent
  identifier, which
• can be used to challenge a peer for its identity
• Does recommendations on content to users
• Has been released on 17 march 2006
• V4.0 released on may 23, 2007
• Has been downloaded about 150,000 times

9
P2P Research in Delft

• People (in PDS)
• Johan Pouwelse
• Henk Sips
• Dick Epema
• Pawel Garbacki
• Michel Meulpolder
• Jan David Mol
• Jie Yang
• Maarten ten Brinke
• Freek Zindel
• Jelle Roozenburg
• Jacco Taal

• Projects
• I-Share
• TLP2PS
• P2P-Fusion
• P2P-Next

• More information
• www.cs.vu.nl/ishare
• www.tribler.org
• www.ewi.pds.tudelft.nl
• (publications database)

Tribler overview paper to appear in
CC-PE, available on www.pds.ewi.tudelft.nl/epem
a
10
Research topic 2 grids

• Grids are large wide-area collections of computer
  resources (supercomputers, clusters, networks,
  storage, etc)
• Usually used for scientific research (simulations
  etc.)
• Problems in grids
• Scheduling and resource management
• Security (authentication, authorization, etc)
• Programming models

11
Research topic 2 grid scheduling

• Problems in grid scheduling
• Grid schedulers usually do not own resources
  themselves
• Grid schedulers have to interface to different
  local schedulers
• The set of grid resources is heterogeneous and
  dynamic
• Workloads are heterogeneous and dynamic
• Structure of applications

load sharing and co-allocation
GS
local schedulers
LS
LS
LS
global job
non-local job
local jobs
12
Grids Research in Delft

• The design and implementation of the KOALA
  co-allocating grid scheduler
• Main goals
• processor co-allocation (un)ordered/flexible
  jobs
• data co-allocation move large input files to the
  locations where the job components will run prior
  to execution
• load sharing in the absence of co-allocation
• run alongside local schedulers
• KOALA
• is written in Java
• uses Globus components (e.g., RSL and GridFTP)
• for launching jobs uses its own mechanisms or
  Globus DUROC
• has been deployed on the DAS2 in september 2005

13
Grids Research in Delft

• People
• Dick Epema
• Alexandru Iosup (grid interoperation)
• Hashim Mohamed (design of KOALA)
• Ozan Sonmez (scheduling)
• Publications
• see PDS publication database at
  www.pds.ewi.tudelft.nl
• Web sites
• www.st.ewi.tudelft.nl/koala
• grenchmark.st.ewi.tudelft.nl
• gwa.ewi.tudelft.nl
• Projects
• VL-e www.vl-e.nl
• GUARD-G, CoreGRID

14
Research topic 3 Wireless sensor networks

• Moores law will bring us
• ultra low-power devices, with
• small form factor, at
• very low cost
• fostering a new range of networked embedded
  systems applications

Smart Dust (1998) Autonomous sensing and
communication in a cubic millimeter
15
WSN The first steps

• Develop COTS hardware
• Develop software (TinyOS)
• Run experiments
• Prototype applications

• ATmega128L CPU (8-bit, 8 MHz)
• 128 KB FLASH (program)
• 4 KB DRAM (data memory)
• Chipcon CC1000 radio (868 MHz)
• modulation FSK 76.8 kBaud
• output power -20 to 10 dBm

LOFAR-agro
Delft/Wageningen
16
WSN Research

• Constraints
• robustness
• limited resources
• energy efficiency

publications

• Topics
• self-configuration
• node localization
• low-bitrate communication
• ad-hoc routing
• in-network data processing
• time synchronization

ACM SenSys, IEEE SECON, EWSN, ACM Trans. on
Sensor Networks,
17
WSN Research in Delft

• People
• Koen Langendoen
• Gertjan Halkes (MAC)
• Aline Baggio (Localization)
• Muneeb Ali (Routing)
• Tom Parker (Data aggr)

• Projects
• LOFAR-agro (precision agriculture)
• Relate (search rescue)
• Smart Surroundings (context awareness)
• Consensus (networking fundamentals)

• IN4181 WSN seminar (4e kw)
• 2 introduction lectures
• 5 seminars (students presenting)
• hands-on lab

• Facilities
• 24 node testbed
• real-time power tracing

18
Research topic 4 parallel systems
high performance embedded systems
19
Research topic 4 Sensors and HPC
LOFAR
processing huge streams of data
20
Parallel Systems Research in Delft

• Projects
• Astrostream
• Scalp

• People
• Henk Sips
• Alexander van Amesfoort
• Rob van Nieuwpoort
• Ana Varbanescu

• Web sites
• astrostream.ewi.tudelft.nl
• scalp.ewi.tudelft.nl

21
PDS Lab facilities the DAS-3

• five clusters
• (one in Delft)
• about 550 CPUs
• 10-Gb lightpaths

There is no end to the computing capacity
required by many applications!!
www.cs.vu.nl/das3
22
PDS MSc Procedure

• Make an appointment with the MSc coordinator
• D.H.J. Epema (tel. 83853, d.h.j.epema_at_tudelft.nl
• Fill out form with list of courses (on EEMCS web
  site)
• PDS colloquium (every 2 months)
• More information on www.pds.ewi.tudelft.nl
• Research projects
• MSc projects (also previous MSc theses)