[PPT] - Rapid Processing of Synthetic Seismograms using Windows Azure Cloud PowerPoint Presentation

SLIDE 1

Rapid Processing of Synthetic Seismograms using Windows Azure Cloud

Vedaprakash Subramanian, Liqiang Wang Department of Computer Science University of Wyoming En-Jui Lee, Po Chen Department of Geology and Geophysics University of Wyoming

1

SLIDE 2

Scientific applications

– Large amount of computing resources – Massive storage for datasets

Traditionally Handled by HPC Clusters

– But Cost-ineffective

Claim: Cloud Computing is a better substitute

2

Introduction

SLIDE 3

Conduct numerical simulation of synthetic

seismograms

Implemented a system on Azure cloud

– to generate synthetic seismograms on the fly based on user queries – to deliver them in real-time

3

… more

SLIDE 4

Seismogram is a record of the ground shaking
Synthetic seismograms are generated by

solving seismic wave – equation

Method is rapid centroid moment tensor

(CMT) inversion method

– Based on 3D velocity model

To increase efficiency, we store Receiver Green

Tensors strain fields

– Generated at the receiver’s location

4

Synthetic Seismograms

SLIDE 5

The input parameters for this wave-equation are

– latitude, longitude and depth of the earthquake locations – strike, dip, and rake (source parameters)

5

… more

SLIDE 6

… more

Helps to map the Earth’s internal structure
Locate and measure the size of different seismic sources
For realistic interpretation of structures
Seismic Hazard Analysis

Why y sy synt nthetic tic se seis ismograms

grams is

is im impor

rtant?

tant?

6

SLIDE 7

Windows Azure

7

SLIDE 8

8

Windows Azure is Platform as a service

architecture

Provides

– Scalable cloud operating system – Data storage system

User controls the hosted application
User cannot control the underlying

infrastructure

Windows Azure

SLIDE 9

Azure service consists of :

– Web role

For web application
For user interface

– Worker role

For generalized development
For background processing
Roles are virtual machines
Say, 2 instances of worker role = 2 virtual

machines running the code of worker role

Service Architecture

SLIDE 10

Blobs
Tables
Queues
Drives
Our system uses only first three

Storage Abstractions

SLIDE 11

Blobs

– Interface for storing files – Two types namely Page blobs and Block blobs – Containers for grouping

Blob Container Account

Geo California

File1 txt File2 txt

Texas

File1 txt

Blob Storage

SLIDE 12

Tables

– Structured storage

– Consists set of entities, which contain a set of properties – Partition key and Row key

Entity Table Account

Geo Customers

Name = … Email = … Name = … Email = …

Photos

Photo Id = … Date = …

Table Storage

SLIDE 13

Queues

– Reliable storage and delivery of messages – Communication between roles

Message Queue Account

Geo Orders

ID = …… ID = …

Queue Storage

SLIDE 14

Azure master system

– Automatically load balance based on the partition key

Partition key for various storage abstractions

– Blobs – Container Name + Blob Name – Entities – Table Name + Partition Key – Messages – Queue Name

14

Load Balancing

SLIDE 15

Implementation of the System

15

SLIDE 16

The architecture of the system :

– Web Role

User interface

– Job Manager

Coordinate the computation
Monitor the system

User Request Queue Computation Input Queue

Windows Azure Storage (Blob, Table, Queue)

Computation Output Queue

Overview of the System

SLIDE 17

– Computation Worker Role

Computation stuff

– 3 Azure Queues

Communication interface between the roles

User Request Queue Computation Input Queue

Windows Azure Storage (Blob, Table, Queue)

Computation Output Queue

… more

SLIDE 18

Receive the user input
Place the request as message in request queue

Web Role

User Request Queue

Windows Azure Storage (Blob, Table, Queue)

SLIDE 19

Retrieve the message from request queue
Read the job
Divide the job into sub-jobs
Place the sub-jobs as message in input-

computation queue

Job Manager

Windows Azure Storage (Blob, Table, Queue)

Request Queue Computation Input Queue Computation Output Queue

SLIDE 20

Num of computation worker roles : 5
Num of CPU cores in each instance : 8
Input : 1000 source locations
Num of queue messages : 1000 / 8 = 125
125 queue messages served by 5 computation

worker roles

Performance gain factor (Theoretical)

= Num of CPU cores * Num of instances = 8 * 5 = 40

Coordinate the Computation

SLIDE 21

Retrieve the message
Process the sub-jobs in parallel using .NET Task

Parallel Library

Write the result
Send message stating job completed

Inside Computation Worker Role

Windows Azure Storage (Blob, Table, Queue)

Computation Input Queue Computation Output Queue

SLIDE 22

Based on the response time of the message
Threshold response time is 2ms
If exceeds

– Allocate a new instance of computational worker role – Maintain its detail

Monitor System Response

SLIDE 23

De-allocate if

– any new instance has been allocated – and its lifetime > one hour – and no message in the queue

… more

SLIDE 24

Distributed File System Computation Input Queue

msg msg msg msg msg

Request to allocate VM

Monitor System Response

SLIDE 25

Allocation & de-allocation are asynchronous
So, mutual exclusion lock are enforced

between

– Job assignment – De-allocation of an instance

Job Manager

SLIDE 26

Seismic wave observations stored as

data file

Each data file is represented by its
wn latitude and longitude
So, blob name = latitude + longitude
For grouping the blobs, the region of

California is divided into blocks

– based on the seismic wave observation stations

Point (40’59’’N, 122’7’’W)

4059-1227

26

Data Storage

SLIDE 27

Currently 4096 stations = 4096

blocks

Each block is characterized by

range of latitude and longitude

So, block identification number =

range of latitude + longitude

Container name = identification

number

3525-3451-1193-11647

34’51’’N, 116’47’’W 35’25’’N, 116’47’’W 34’51’’N, 119’3’’W 35’25’’N, 119’3’’W 27

… more

SLIDE 28

Divide California region

into 16 parts

1 Table for 1 part
Store list of blocks in the

part into the table

Helps in better retrieval

28

… more

SLIDE 29

Locate blob corresponding to the given point

– Retrieve the container name and blob name

Retrieve container name from the table

– Locate the table – Do a linear search inside the table

Blob name = latitude + longitude

29

Data Query Algorithm

SLIDE 30

Point 41’43’’N, 124’6’’W 41’43’’N, 120’43’’W 39’11’’N, 120’43’’W 39’11’’N, 124’6’’W 40’27’’N, 124’6’’W 39’11’’N, 122’52’’W 40’27’’N, 122’52’’W 40’27’’N, 123’56’’W 40’85’’N, 122’52’’W 40’85’’N, 123’56’’W 40’59’’N, 122’7’’W 30

Data Query Algorithm

SLIDE 31

Experiment

31

SLIDE 32

Performance was evaluated on

– various configurations and number of instances of computational worker role – datasets from different number of seismic wave

bservation

32

Experiment

SLIDE 33

100 200 300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1000 Single Worker (4 core) Four Worker (4 core) Single worker (4 core) + TPL Four Worker (4 core) + TPL Two worker (8 core) + TPL Number of stations Execution Time (seconds)

33

Performance Measurement – Execution Time

SLIDE 34

Conclusion

34

SLIDE 35

Implemented the system on Windows Azure
Hence Cloud is a better substitute
Future Work :

– Add Seismic Hazard Analysis feature to the system

35

Conclusion

SLIDE 36

Thanks

36