Data Grids Data Intensive Computing

1
Data GridsData Intensive Computing
2
Simplistically

• Data Grids
• Large number of users
• Large volume of data
• Large computational task involved
• Connecting resources through a network

3
Grid

• Term Grid borrowed from electrical grid
• Users obtains computing power through Internet by
  using Grid just like electrical power from any
  wall socket

4
Data Grid

• By connecting to a Grid, can get
• needed computing power
• storage spaces and data
• Specialized equipment
• Each user - a single login account to access all
  resources
• Resources - owned by diverse organizations
  Virtual Organization

5
Data Grids

• Data
• Measured in terabytes and soon petabytes
• Also geographically distributed
• Researchers
• Access and analyze data
• Sophisticated, computationally expensive
• Geographically distributed
• Queries
• Require management of caches, data transfer over
  WANs,
• Schedule data transfer and computation
• Performance estimates to select replicas

6
Data Grids

• Domains as diverse as
• Global climate change
• High energy physics
• Computational genomics
• Biomedical applications
• http//www.gridstart.org/links.shtml

7
Data Grids

• Data grid differs from
• Cluster computing grid is more than homogeneous
  sites connected by LAN (grid can be multiple
  clusters)
• Distributed system grid is more than
  distributing the load of a program across two or
  more processes
• Parallel computing grid is more than a single
  task on multiple machines
• Data grid is
• heterogeneous, geographically distributed,
  independent site
• Gridware manages the resources for Grids

8
Methods of Grid Computing

• Distributed Supercomputing
• Tackle problems that cannot be solved on a single
  system
• High-Throughput Computing
• goal of putting unused processor cycles to work
  on loosely coupled, independent tasks (SETI
  Search for Extraterrestrial Intelligence)
• On-Demand Computing
• short-term requirements for resources that are
  not locally accessible, real-time demands

9
Methods of Grid Computing

• Data-Intensive Computing
• Synthesize new information from data that is
  maintained in geographically distributed
  repositories, databases, etc.
• Collaborative Computing
• enabling and enhancing human-to-human
  interactions

10
An Illustrative Example

• NASA research scientist
• collected microbiological samples in the
  tidewaters around Wallops Island, Virginia.
• Needs
• high-performance microscope at National Center
  for Microscopy and Imaging Research (NCMIR),
  University of California, San Diego.

11
Example (continued)

• Samples sent to San Diego and used NPACIs
  Telescience Grid and NASAs Information Power
  Grid (IPG) to view and control the output of the
  microscope from her desk on Wallops Island.
• Viewed the samples, and move platform holding
  them, making adjustments to the microscope.

12
Example (continued)

• The microscope produced a huge dataset of images
• This dataset was stored using a storage resource
  broker on NASAs IPG
• Scientist was able to run algorithms on this very
  dataset while watching the results in real time

13
Grid topology
Grid Topology
14
Higher level services

• Replica selection
• Choosing specific replica to optimize
  performance, cost, security
• Grid info services provide info about network,
  metadata provide info about size of file
• Determine replica with fastest access and/or
  determine of replicate will result in better
  access
• Subsets of files

15
Scheduling and Replication in Data-Intensive
Computing

• Ming Lei, PhD student
• Department of Computer Science
• University of Alabama

16
Introduction

• Early work in data replication focused on
  decreasing access latency and network bandwidth
• As bandwidth and computing capacity become
  cheaper, data access latency can drop
• How to improve availability and reliability
  becomes the focus
• Due to dynamic nature of Grid user/system,
  difficult to make replica decisions to meet
  system availability goal
• Usually assumed unlimited storage

17
Introduction

• Related work
• Economical model replica decision based on
  auction protocol Carman, Zini, et al. e.g.,
  replicate if used in future, unlimited storage
• Hotzone places replicas so client-to-replica
  latency minimized Szymaniak et al.
• Replica strategies dynamic, shortest
  turnaround, least relative load Tang et al.
  consider only LRU
• Multi-tiered Grid Simple Bottom Up and
  Aggregate Bottom Up Tang et al.
• Replicate fragments of files, block mapping
  procedure for direct user access ChangChen

18
Sliding Window Replica Scheme

• Alternative to future prediction which can
  overemphasize future accesses times when queue is
  long
• Following example based on future access times

19
Figure 2. Without Replica File 4
Figure 3. Replica File 5
20
Sliding window replica protocol

• Build sliding window set of times used
  immediately in the future
• Size bound by size of local SE
• Includes all files current job will access and
  distinct files from next arriving jobs
• Sliding window slides forward one more file each
  time system finishes processing a file
• Sliding window dynamic

21
Sliding window replica protocol

• Sum of size of all files lt size of SE
• No duplicate files in sliding window

22
(No Transcript)
23
OptorSim

• Evaluate the performance of our replica and
  replacement strategy
• Using OptorSim
• OptorSim developed by the EU DataGrid Project to
  test dynamic replica schemes

24
Simulation results

• Assume OptorSim topology
• 10,000 jobs in Grid
• Each job accesses 3-10 files
• Storage available 100M-10G
• File size 0.5-1.5 G
• Compare replica strategies
• LRU, LFU, EcoBio, EcoZipf, No Prediction, Sliding
  Window

25
Measurements

• Measurements
• Total Running time

26
Eco model

• Compare to Economical Model in OptorSim
• Eco
• File replicated if maximizes profit if SE (e.g.
  what is earned over time based on predicted
  number of file requests)
• Eco prediction functions
• EcoBio
• EcoZipf
• Queue Prediction
• No Prediction

27
Prediction Functions

• Prediction via four kinds of prediction
  functions
• Bio Prediction binomial distribution is used to
  predict a value based on file access history
• Zipf Prediction Zipf distribution is used to
  predict a value based on file access history
• Queue Prediction The current job queue is used
  to predict a value of the file
• No Prediction No predictions of the file are
  made, the value will always be 1

28
Access Patterns

• Consider 4 access patterns
• Random
• Random Walk Gaussian
• Sequential
• Random Walk Zipf

29
Running Time

• First, study sliding window without RDAE
• Measure running time

30
Figure 7. Running Time with Varying File
Accessing Pattern.
31
Sliding Window

• Sliding window replica scheme always best
  turnaround time
• No replication, EcoBio the worst
• LFU second best

32
Figure 8. Impact of network bandwidth
33
Sliding Window

• Higher the bandwidth, shorter system performance
• Sliding window replica always the best
• EcoZipf, EcoBio perform almost same as no
  prediction

34
Switch Time

• Impact of switch time

35
Figure 10. Impact of varying the job switch time
36
Switch Time

• Longer the switch time, the longer the total
  running time
• Sliding window the best
• LRU, LFU second best
• Improvement provided by sliding window over LRU,
  LFU greatest for smaller switch times
• Improvement provided by sliding window over
  EcoBio, EcoZipf greatest remains high for higher
  switch times

37
Varying Schedulers

• Impact of varying schedulers

38
Figure 11. Running Time with Varying Schedulers
39
Conclusions and Future work

• File bundle situation
• Preferential treatment of smaller size files
• Green Grids

40
Power Aware

• Info from Pate, et al., 2002
• Data center with 1000 racks, 25,000 square feet
  require 10 MW of power.
• Also requires 5 MW to dissipate the heat
• At 100/MWh, 4M per year
• Average utilization of 20 and 80 for peak loads
• This means 80 of resources not utilized, yet
  generating heat, consuming power

41
Power Aware

• Green Grid group of IT professionals
• Power Usage Effectiveness PUE
• Total facility power/IT equipment power
• Data Center infrastructure Efficiency metric CDiE
• 1/PUE

42
Power Aware

• Computer Dec. 2007 devoted to green computing
• Servers
• Seek high energy efficiency at peak performance
• In sleep consume near-zero energy
• But
• Servers rarely completely idle, seldom operate
  near maximum utilization
• Operate at 10 to 50 of maximum utilization

43
Power Aware

• Mismatch between server energy efficiency and
  behavior of server class workloads
• Need energy proportional machines to exhibit wide
  dynamic power range

44
What can be done to achieve energy proportional
behavior

• Server power (Google) peak vs. idle
• CPU smaller and smaller fraction of total power
  when system idle
• Processors close to energy proportional behavior
• Desktop and server processors can consume less
  than 1/3 of their peak power at low activity
  modes can achieve 1/10 or less
• Dynamic power range of other components much
  narrower
• Lower voltage processor frequency mode good
  because not much impact on overall performance
• Networking equipment doesnt offer low power
  modes
• Need improvements in memory and disk systems

45
Future Wwork - Power Aware

• Ways to conserve energy
• Power down CPU
• Slow down CPU
• Power down storage disks energy to power up
• Store less data

46
Future Work

• With Dr. John Lusth
• Build a green grid (with lower energy mother
  boards)
• Hardware purchased, put together
• Implement some power saving strategies
• Run benchmarks (e.g. computations, queries)
• Compare to existing grid at Arkansas
• DESIGN NEW STRATEGIES