energy efficient prefetching – from models to implementation 6/19/2015 1 adam manzanares and xiao...

49
Energy Efficient Prefetching – from models to Implementation 06/20/22 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering Auburn University http://www.eng.auburn.edu/~xqin [email protected]

Post on 20-Dec-2015

216 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Energy Efficient Prefetching – from models to Implementation

04/18/23 1

Adam Manzanares and Xiao Qin

Department of Computer Science and Software Engineering

Auburn Universityhttp://www.eng.auburn.edu/~xqin

[email protected]

Page 2: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Adam Manzanares

Ph.D. May 2010.

Page 3: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

About me

Ph.D.’04, U. of Nebraska-Lincoln

04-07, New Mexico Tech 07-10, Auburn University

Page 4: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

About My Research Group

Page 5: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Presentation Outline

• Motivation

• Modeling Work

• DiskSim Modifications

• Energy Efficient Virtual File System (EEVFS)

• Parallel Striping Groups in EEVFS

• Conclusion

04/18/23 5

Page 6: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

MotivationEPA Report to Congress on Server and Data Center Energy Efficiency, 2007

04/18/23 6

Page 7: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Motivation Using 2010 Historical Trends Scenario

◦ Server and Data Centers Consume 110 Billion kWh per year

◦ Assume average commercial end user is charged 9.46 kWh

◦ Disk systems can account for 27% of the energy cost of data centers

04/18/23 7

Page 8: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Buffer Disk Architecture

RAM BufferRAM Buffer

m buffer disksm buffer disks n data disksn data disks

Buffer Disk ControllerBuffer Disk Controller

Data Partitioning Data Partitioning

Security Model Security Model

Load BalancingLoad Balancing

Power ManagementPower Management

PrefetchingPrefetching

Disk RequestsDisk Requests

Energy-Related Reliability Model Energy-Related Reliability Model

04/18/23 8

Page 9: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

IBM Ultrastar 36Z15

04/18/23 9

Transfer Rate 55 MB/s Spin Down Time: TD 1.5 s

Active Power: PA 13.5 W Spin Up Time: TU 10.9 s

Idle Power: PI 10.2 W Spin Down Energy: ED 13 J

Standby Power: PA 2.5 W Spin Up Energy: EU 135 J

Break-Even Time: TBE 15.2 S

Page 10: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Prefetching

Disk 1

Disk 2

Disk 3

Buffer Disk

04/18/23 10

Page 11: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Why Modeling & Simulation

• Allows us to determine the potential of our research ideas

• Can quickly evaluate many simulation parameters

• Allows us to test architectures and hardware without having the physical resources

04/18/23 11

Page 12: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Modeling & Simulation Work

Developed Mathematical Model◦Disk Energy Consumption◦Conditions to prefetch

Developed Energy Saving Principles◦Investigated cases that exploit the energy

saving principles Implemented model in JAVA based simulator

04/18/23 12

Page 13: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Energy Saving Principles

Energy Saving Principle One◦Increase the length and number of idle

periods larger than the disk break-even time TBE

Energy Saving Principle Two◦Reduce the number of power-state

transitions

04/18/23 13

Page 14: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Paramaters TestedParameter Values

Data Size 1, 5, 10, 25 MB

# of Data Disks 4, 8, 12

Inter-arrival Delay 0, 0.1, 0.5, 1 S

Hit Rate 85, 90, 95, 100%

04/18/23 14

Page 15: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Energy Savings Hit Rate 85%

04/18/23 15

Page 16: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

State Transitions

04/18/23 16

Page 17: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Parameter Generalizations

• Larger data sizes produce greater energy savings and less state transitions

• Increasing the inter-arrival delay increases energy savings

• More data disks per buffer disks increases energy efficiency

• High hit rates produce the greatest energy efficiency

04/18/23 17

Page 18: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Modeling & Sim. Summary

Hit Rate, Inter-arrival Delay, & Data Size combine to produce Idle Windows

Transitions important to reduce energy consumption◦ May increase/decrease to reduce energy consumption

Disk parameters have large impact on energy savings

Model and simulator developed in-house

04/18/23 18

Page 19: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

DiskSim

• Event driven simulator developed at CMU

• Simulates disks at the block level

• The simulator has been validated

• Discrete event based simulator

• Provides a large amount of statistics

• Lacks Disk Power Models

• Ability to simulate large storage systems

04/18/23 19

Page 20: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

File System Simulator

• Large files important to energy savings

• Popularity of data is also useful

• Developed a block to file translator

• Interacts with DiskSim

04/18/23 20

Page 21: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

DiskSim with File System Simulator

04/18/23 21

Page 22: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Modified DiskSim Results

04/18/23 22

Page 23: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Modified DiskSim Summary

• Provides us with accurate disk statistics

• Only the changes to DiskSim need to be validated

• Heavily dependent upon disk parameters

• May miss details that can only be found in implementation

04/18/23 23

Page 24: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Why a Cluster File System

• Block level prefetching difficult

• Natural place to track file accesses

• Control placement of data among storage nodes, and data disks

• Tiered approach simplifies management of files and disk states

• Eliminates some shortcomings of modeling and simulation

04/18/23 24

Page 25: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Energy Efficient Virtual File System

04/18/23 25

Page 26: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

EEVFS Process Flow

04/18/23 26

Page 27: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

EEVFS Testbed

Parameter Storage Server Storage Node Type 1

Storage Node Type 2

CPU P4 2.0 GHz P4 3.2 GHz P4 2.4 GHz

Memory (MB) 2000 1000 512

Network Interconnect

1000 1000 100

Disk Type SATA ATA/133 ATA/133

Disk Capacity 120 GB 80 GB 80 GB

Disk Bandwidth 100 MB/s 58 MB/s 34 MB/s

04/18/23 27

Page 28: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Energy Savings

04/18/23 28

Page 29: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

State Transitions

04/18/23 29

Page 30: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Response Times

04/18/23 30

Page 31: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Berkeley Web Trace

04/18/23 31

Page 32: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

EEVFS Summary

• Knowledge of requests assumed and may be hard to come by

• Performance tied to one of the buffer disks

04/18/23 32

Page 33: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Parallel Striping Groups

Disk 1 Disk 2

Group 1

Buffer Disk

Storage Node 1

Disk 3 Disk 4Buffer Disk

Storage Node 2

Disk 5 Disk 6

Group 2

Buffer Disk

Storage Node 3

Disk 7 Disk 8Buffer Disk

Storage Node 4

File 1 File 2File 3 File 4

04/18/23 33

Page 34: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Striping Within a Group

Disk 1 Disk 2

Group 1

Buffer Disk

Storage Node 1

Disk 3 Disk 4Buffer Disk

Storage Node 2

1 3 5 7 9 4 6 8

4 6 81 3 5 7 9

10

10

1

2

1

2

File 1 File 22 2

04/18/23 34

Page 35: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Striping Within a Group

• Number of disks in a group can be matched to nearest bottleneck

• Striping within the group maintains relatively high performance

• Allows us to use a buffer disk for each storage node, while still maintaining file striping level

04/18/23 35

Page 36: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

TestbedParameter Storage Server Storage Node

CPU Celeron 2.2 GHz Celeron 2.2 GHz

Memory (MB) 2000 2000

Network Interconnect

1000 1000

Disk Type SATA SATA

Disk Capacity 160 GB 480 GB

Disk Bandwidth 126 MB/s 126 MB/s

04/18/23 36

Page 37: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Measured Results

04/18/23 37

Page 38: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Measured Results

04/18/23 38

Page 39: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Berkeley Web Trace

04/18/23 39

Page 40: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Response Time Comparison

• Energy efficiency is slightly improved

• Response time gain is significant

Parameter Striping No Striping

Energy Consumption (J) 2,088,113 2,100,243

Response Time (S) 2.78 13.87

04/18/23 40

Page 41: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Parallel Striping Groups Summary

• Improves the energy efficiency and performance of a storage system

• Designed to scale– Needs to be tested on large scale storage

system

04/18/23 41

Page 42: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Conclusions

• Modeling and simulation used to test our ideas

– System, Disk, Trace Parameters varied to study their impacts

• DiskSim Modifications

– Added disk power models to DiskSim

– Implemented block to file translator

• Energy Aware Virtual Cluster File System (EEVFS)

– Implemented a prototype

– Added parallel striping groups to improve the energy efficiency

04/18/23 42

Page 43: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Future Work

• Improve the EEVFS prototype for production use

• Run EEVFS on large scale storage system– Investigate scaling effects

04/18/23 43

Page 44: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

http://www.auburn.edu/~xzq0001

Page 45: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Download the presentation slides

Page 46: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Download the presentation slides

Page 47: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Download the presentation slides

Page 48: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

http://www.slideshare.net/xqin74

Page 49: Energy Efficient Prefetching – from models to Implementation 6/19/2015 1 Adam Manzanares and Xiao Qin Department of Computer Science and Software Engineering

Questions