Sie sind auf Seite 1von 17

Majd F.

Sakr
msakr@cs.cmu.edu

Location

In-Class

Online

Scale

10s
students

>100
students

Campus

Local

Global

Infrastructure

Local

Outsourced

Systems &
Infrastructures

Course
Content

Services &
Applications

CMU-SV

CMU-P
CMU-Q
CMU-R

Undergrad
9 units
(10%)

CMU-A

330
Graduate
12 units
90%

170

213

50
Spring 2013

Fall 2013

Spring 2014

Fall 2014

Amazon Web
Services

Cloud platform to
implement all projects

Open Learning
Initiative

Piazza

Feedback
& questions

Access learning content,


quizzes and checkpoints

Video Conf.

Weekly, in-class recitations


videotape & video conference

Real world data, infra, problems within budget

0. AWS Account Setup & Tool Primer

Amazon EC2, Amazon S3, SSH,


Authentication, Billing, Security Groups

1. Big Data Analytics

Amazon EC2, Amazon EMR

2. Scaling, Elasticity and Failure

Amazon EC2, Auto Scaling, ELB,


Amazon CloudWatch, Amazon SNS

3. Cloud Storage

Amazon EBS, MySQL, HBase,


Amazon RDS, Amazon DynamoDB

4. Analytics Engines for the Cloud

Amazon EMR, MapReduce, Spark,


GraphLab

5. A Complete Web Service (Team project)

No restrictions on tools or AWS services


Evaluated based on cost and performance

Student details

Database
(RDS)
Update

Cron job to
update DB

Authentication
Server

Auto-grader Server
AWS ID,
Instance ID,
AMI ID
Submitter

Andrew ID,
Link to update script
(Or 0 if not authenticated)

Submission token

Student VM
Instance

Runner.sh (local)
Submitter (remote)

Fetch update script, run it and delete

Student details

Database
(RDS)
update DB
Update

Load-generator &
Auto-grader

Load

Student Submits IP of Web Service

Authentication
Server

AWS ID,
Instance ID,
AMI ID

Response

Student Front End VM


of Web Service

Andrew ID,
Link to update script
(Or 0 if not authenticated)

Load
Generator

Phases

Duration

Query Type

Phase 1

2 weeks

Q1, Q2

Phase 2

2 weeks

Q1, Q2, Q3, Q4

Phase 2 Live Test

6 hours

Q1, Q2, Q3, Q4, mix-Q1Q2Q3Q4

Phase 3

2 weeks

Q1, Q2, Q3, Q4, Q5, Q6

Phase 3 Live Test

6 hours

Q1, Q2, Q3, Q4, Q5, Q6, MIX-Q1Q2Q3Q4Q5Q6

Testing

Develop idea

Identify data set


Design project steps
Develop project pilot
Compare results with insight

V1.0 triple tested


V2.0 released

Bugs

Develop project learning


outcomes
Understand capability of
AWS services

Design &
Development

Data set size


Estimate student effort
Types of resources
Number of resources

Tradeoff scale
with cost

Auto Patching of
AMIs

Evaluation

Project questions
Bash logs
AWS expenditure
Code submission

Cheat Checking
Code similarity check
using MOSS
Correlation across multiple logs

Bash
Logs

AWS
Cost

Eval

Monitor Expenditure
Daily
Per project module (tagging)

Code
on S3

Project
Qs

Cheating should be more work than learning

Full Score = Correctness + Performance + Budget

F14: 10% penalties: 43 in Week 1 0 in Week 8


F14: 100% penalties: 6 in Week 1 0 in Week 8
S15: 1 penalty in weeks 1-3