4/10/2012. Hall. Cherokee. Bartow. Forsyth. Barrow. Gwinnett. Paulding. Cobb. Walton. DeKalb. Douglas. Fulton. Rockdale. Carroll. Clayton.



Similar documents
Setting Up the ARC Activity-Based Model in the Cloud: Lessons to Date. Ben Stabler, PB. Guy Rousseau, ARC. Matthew Martimo, Citilabs

Who Has (And Doesn t Have) Health Insurance in Metro Atlanta?

Average Energy Prices, Atlanta September 2015

45-64 AGE GROUP IS FASTEST GROWING SEGMENT

Individual and Family Plans Georgia Kaiser Permanente Plan Highlights

2014 Outline of Medicare Supplement Coverage Cover Page (1 of 2) Plans A, F & N

Appendix: Economic Opportunity

2015 Outline of Medicare Supplement Coverage Cover Page (1 of 2) Plans A, F & N

Cloud Computing and Amazon Web Services

CloudFTP: A free Storage Cloud

Genworth 2015 Cost of Care Survey Georgia

THE NAIS DEMOGRAPHIC CENTER Metropolitan Area Reports

Atlanta Journal-Constitution

HIV Surveillance Update

Open Cloud System. (Integration of Eucalyptus, Hadoop and AppScale into deployment of University Private Cloud)

Clearing Away the Clouds: Cloud Computing at Stanford

Hardware/Software Guidelines

Atlanta South Division

Cloud computing is a marketing term that means different things to different people. In this presentation, we look at the pros and cons of using

ST 810, Advanced computing

Visualisation in the Google Cloud

CLOUD PERFORMANCE TESTING - KEY CONSIDERATIONS (COMPLETE ANALYSIS USING RETAIL APPLICATION TEST DATA)

jbase 5 Install on Amazon AWS a Primer

Technical Aspects to GIS in the Cloud

For 6 months ended September 30, 2010 ATLANTA GROUP

Using SUSE Studio to Build and Deploy Applications on Amazon EC2. Guide. Solution Guide Cloud Computing.

Liferay Portal Performance. Benchmark Study of Liferay Portal Enterprise Edition

Fixed Price Website Load Testing

GEORGIA DEPARTMENT OF COMMUNITY AFFAIRS OFFICE OF RESEARCH 60 EXECUTIVE PARK SOUTH ATLANTA, GEORGIA

ECE6130 Grid and Cloud Computing

Paulding County School District 2015 BENEFITS OPEN ENROLLMENT REVIEW

Configuration Maximums VMware Infrastructure 3

On Demand Satellite Image Processing

Server Installation Manual 4.4.1

Planning Level Cost Estimation Tool. User s Manual

WHITE PAPER SETTING UP AND USING ESTATE MASTER ON THE CLOUD INTRODUCTION

Data Centers and Cloud Computing

Dedicated Hosting. The best of all worlds. Build your server to deliver just what you want. For more information visit: imcloudservices.com.

2015 State Health Benefit Plan Annual Open Enrollment (OE) Presentation to Active Employees and pre-65 Retirees

Where We Are. References. Cloud Computing. Levels of Service. Cloud Computing History. Introduction to Data Management CSE 344

Amazon EC2 XenApp Scalability Analysis

LabStats 5 System Requirements

Migration Scenario: Migrating Batch Processes to the AWS Cloud

Description of Application

Secure Cloud Computing with FlexCloud

System requirements. for Installation of LANDESK Service Desk Clarita-Bernhard-Str. 25 D Muenchen. Magelan GmbH

CLOUD. MADE EASY. vnebula Portal

Cloud Computing. Adam Barker

A programming model in Cloud: MapReduce

vnebula Cloud. Made Easy. Introducing vnebula from Stream Networks. A simple, self-service cloud portal for our partner community.

G ULF C OAST W ORKFORCE B OARD

PostgreSQL Performance Characteristics on Joyent and Amazon EC2

2016 AP STEM ACHIEVEMENT SCHOOLS

Manjrasoft Market Oriented Cloud Computing Platform

2016 AP HUMANITIES SCHOOLS

Cloud Computing For Bioinformatics

Submit to: PT. Adi Sarana Armada (Assa Rent) Proof of Concept (POC) of 2X Solution. Ver. 1.0 Mei Prepared by: PT. Netmarks Indonesia

A PERFORMANCE ANALYSIS of HADOOP CLUSTERS in OPENSTACK CLOUD and in REAL SYSTEM

System requirements for MuseumPlus and emuseumplus

Virtual computers and virtual data storage

GeoCloud Project Report USGS/EROS Spatial Data Warehouse Project

White Paper: Deploying QlikView

FREE computing using Amazon EC2

INTRODUCTION TO WINDOWS 7

How To Image A Single Vm For Forensic Analysis On Vmwarehouse.Com

A Web Base Information System Using Cloud Computing

Installing and Configuring Windows Server Module Overview 14/05/2013. Lesson 1: Planning Windows Server 2008 Installation.

Security Benefits of Cloud Computing

Virtualization and Cloud Computing

Session 3. the Cloud Stack, SaaS, PaaS, IaaS

Data Centers and Cloud Computing. Data Centers. MGHPCC Data Center. Inside a Data Center

AWS Account Setup and Services Overview

ArcGIS for Server in the Amazon Cloud. Michele Lundeen Esri

SEAIP 2009 Presentation

Comparison of Cloud vs. Tape Backup Performance and Costs with Oracle Database

Server Software Installation Guide

Cloud Computing and E-Commerce

Cloud Computing: Computing as a Service. Prof. Daivashala Deshmukh Maharashtra Institute of Technology, Aurangabad

Copyright 2012 California Institute of Technology. Government sponsorship acknowledged. ( )

Citrix XenApp-7.6 Administration Training. Course

Getting Started with HC SharePoint Module

Running R from Amazon's Elastic Compute Cloud

Matchmaking in the Cloud: Amazon EC2 and Apache Hadoop at eharmony

Cloud Computing: A Brief Summary. Lucid Communications Limited Prepared by Neil Turner September 2009

Jeffrey D. Ullman slides. MapReduce for data intensive computing

Unlimited Virtual Computing Capacity using the Cloud for Automated Parameter Estimation

Data Centers and Cloud Computing. Data Centers

Putchong Uthayopas, Kasetsart University

International Journal of Engineering Research & Management Technology

Cloud Computing & Hosting Solutions

QHR Accuro EMR IT Hardware Requirements

Introduction to Cube Cloud (CC) Introduction. Senior Transportation Engineer

GTC Presentation March 19, Copyright 2012 Penguin Computing, Inc. All rights reserved

wu.cloud: Insights Gained from Operating a Private Cloud System

Use of Cloud Computing for scalable geospatial data processing and access

Research Data Storage, Sharing, and Transfer Options

Het is een kleine stap naar een hybrid cloud

DOCLITE: DOCKER CONTAINER-BASED LIGHTWEIGHT BENCHMARKING ON THE CLOUD

Pure Cloud Run all components of the Apica ProxySniffer product completely in the cloud.

InstaFile. Complete Document management System

Transcription:

Carroll Bartow Paulding Douglas Coweta Cobb Cherokee Fulton Fayette Clayton DeKalb Spalding Forsyth Henry Gwinnett Rockdale Hall Newton Barrow Walton 4/10/2012 Cloud Computing On demand use of remote computer resources Examples: Yahoo Mail Google Apps (such as Calendar, Docs, etc) Cube Cloud Controller Amazon Elastic Compute Cloud (Ec2) Key characteristics of cloud computing: Scalable Fee structure usually on demand or subscription based Supports multiple users/instances 1 3 Topics Is Cloud Computing the way to go? ARC ABM Review Configuration Basics Setting up the ARC Cloud Based ABM Hardware Configuration Software Configuration Run Times Inputs and Outputs Client User Interface Future Work Questions 2 ARC ABM Review Based on the CT RAMP family of ABMs developed, or being developed, in Columbus, Atlanta, the San Francisco Bay Area, San Diego, Phoenix, Chicago, Miami, and other regions ARC model implemented with Cube based networks, assignments and skimming, GUI and ancillary models (external model, truck model, etc) Population Size: 1.7 million households in 2005, 2.7 million in 2030 Uses Java Parallel Processing Framework and Cube Cluster to thread and distribute work across multiple machines Base year runs 3 feedback loops in 16 hours on the ARC cluster using 3 Windows 64bit machines with 8 processors and 32 GB of RAM each 4 1

ARC ABM System Architecture Select and Configure Machine Instances Standard Instances (1.7GB RAM, 1 Core, 160GB HD, 32bit) Large Instances (7.5GB RAM, 4 Core, 850GB HD, 64bit) Extra Large Instances (15GB RAM, 8 Core, 1690GB HD, 64bit) High Memory Extra Large Instances (17GB RAM, 6.5 Core, 420GB HD, 64bit) High Memory Double Extra Large Instances (34GB RAM, 13 Core, 850GB HD, 64bit) High Memory Quadruple Extra Large Instances (68GB RAM, 26 Core, 1690GB HD, 64bit) Cluster Compute Quadruple Extra Large Instances (23GB RAM, 33.5 Core, 1690GB HD, 64bit, 10GBit Ethernet) Windows Server or various Linux versions Various database options and web server options as well Estimate overall price tag: http://aws.amazon.com/calculator 5 7 Configuration Basics Goal: Setup the ABM in the cloud and see how it performs Two basic approaches to the cloud based setup: System is open so the user can configure and use custom remote machines Example: rent a few instances (i.e. remote machines) from Amazon E2C and configure them as needed Advantage: flexibility Disadvantage: complexity System is configured ahead of time and a client user interface is developed that sits between the user and the remote machines Example: Cube Cloud Controller, which wraps a user interface around Amazon EC2 and allows the user to upload files, run models, and get results Advantage: ease of use Disadvantage: lack of flexibility On Demand Instance Pricing 6 8 2

Reserved Instance Pricing 9 Manage your instances with Management Console Remote desktop to instances and install Java and Cube We used Amazon EC2 instances configured by Citilabs so they already had a Cube site license (no hardware key version) installed on them However, most modeling packages require a hardware key This is not possible with Amazon EC2 Possible future solutions to this include: Remote license checkout Purchasing a one time use key for each model run instead of purchasing an unlimited use hardware key The cloud machines are only used for running the model, and network editing is still done on a local machine 11 ARC ABM Example Computing Cost (which does not include data transfer costs) Configure 3 High Memory Double Extra Large Instances On Demand 16 hour model run $60 Reserved Instances for 3 years Initial fee $12,000 16 hour model run $23 Need to run the model 324 times in 3 years to justify initial fee 10 Run Times Two runs tested at this point: 3 machines with 6 feedback loops 6 machines with 6 feedback loops The 3 machine cloud run time is comparable to the ARC cluster run time The move from 3to 6 machines resulted in a ~35% reduction in run times More test runs are planned, including with more instances, as well as with some performance tuning Run Times (hh:mm) Iteration 3 machine 6 machine 1 3:11 3:03 2 6:11 2:50 3 7:38 4:07 4 7:16 4:37 5 7:32 5:04 6 11:37 7:30 Total Java Run Time 43:25 27:11 Total Model Run Time 48:10 31:25 12 3

Another Appproach More model runs (32 cores machines, 64, 128, 256, 512): Increase in computing instances resulted in significant nonlinear reductions in the ABM run time. Doubling the number of cores from 32 to 64 reduced the run time by 37 percent. Doubling it again reduced the run time relative to the 32 core run by 55 percent. The 256 and 512 core runs show little improvement beyond the 128 core run. ARC ABM computing power sweet spot is somewhere around 128 cores. Inputs and Outputs ARC ABM Example Data Transfer Cost 250MB for zipped complete model setup $0.025 However, only key inputs for a scenario are likely to be uploaded so the cost is usually less 15 GB of output for a compute model run $2.25 However, only key outputs are likely to be downloaded so the cost is usually less 13 15 Inputs and Outputs Separate cost to transfer data in and out of Amazon s Ec2 cloud Client User Interface Amazon EC2 is designed for software developers so it is not really that easy to use As a result, a client web based user interface would need to be developed that allows the user to easily use the cloud based ABM An example client user interface is Cube Cloud Controller, which allows users to: Upload model inputs Run and manage scenarios Download model outputs 14 16 4

Future Work Experiment with Cube Cloud Controller for managing runs in the cloud Ease of use Cost Multiple user capabilities Develop system to allow multiple users to run the ABM simultaneously or in a more coordinated fashion we need a cluster is busy light ARC review the costs of in house runs versus cloud runs and develop a work program that is most efficient for the agency 17 5