computing at scale

25
Computing at Scale: Data Exploration Jerjou Cheng, Barry Brumitt

Upload: jerjou

Post on 27-Jan-2015

114 views

Category:

Technology


3 download

DESCRIPTION

Talk given at Seattle Tech Forum on Dev 15, 2010 at Bellevue City Hall.

TRANSCRIPT

Page 1: Computing at scale

Computing at Scale:Data Exploration

Jerjou Cheng, Barry Brumitt

Page 2: Computing at scale

HELLO

/jur'jō/ɯoɔ˙ǝlƃooƃ@noɾɹǝɾ

MY NAME IS

Developer Programs Engineer(Developer Relations)

Google Storage

Page 3: Computing at scale

Michael PIERROThttp://www.freephotobank.org/v/sky-stars/cloud/Cloud-19.jpg.html

Page 4: Computing at scale

Computing at Scale:Data Exploration

Jerjou Cheng, Barry Brumitt

Page 5: Computing at scale

Overview

• Google App Engine• Google Storage for Developers• Prediction API• BigQuery

Page 6: Computing at scale

Introductions

Page 7: Computing at scale

Who are theseservices for?

Page 8: Computing at scale

A World without Clouds

• Build a web applicationoStartup costsoMaintenance / reliabilityoScaling

Michael Scheltgenflickr.com/mscheltgen/

Page 9: Computing at scale

Google App Engine

Easy to startEasy to maintainEasy to scale

Page 10: Computing at scale

Users

Page 11: Computing at scale

gigy Socialize - traffic

Page 12: Computing at scale

Overview

• Google App Engine• Google Storage for Developers• Prediction API• BigQuery

Page 13: Computing at scale

Overview

• Google App Engine• Google Storage for Developers• Prediction API• BigQuery

Page 14: Computing at scale

A World without Clouds

• Store data• Reliability• Sharing• Large objects

Michael Scheltgenflickr.com/mscheltgen/

Page 15: Computing at scale

Google Storage for Developers

• Google infrastructure• You control access to your data• Store massive data in Google's cloud• Easy interface

Page 16: Computing at scale

Example

 

Page 17: Computing at scale

Internal use cases

• Content hosting  

  • Sharing

  

• Data Import  Google BigQuery

Google Prediction API

Page 18: Computing at scale

Overview

• Google App Engine• Google Storage for Developers• Prediction API• BigQuery

Page 19: Computing at scale

Prediction API

• Cloud-hosted machine learning as service• Simple interface over complex analysis• Predict results in real-time

Page 20: Computing at scale

How does it work?

"english" The quick brown fox jumped over the lazy dog.

"english" To err is human, but to really foul things up you need a computer.

"spanish" No hay mal que por bien no venga.

"spanish" La tercera es la vencida.

"english" To be or not to be, that is the question.

"spanish" La fe mueve montañas.

The Prediction APIfinds relevantfeatures in the sample data during training.

The Prediction APIlater searches forthose featuresduring prediction.

Page 21: Computing at scale

Prediction API

1. Upload

2. Train

Upload your training data toGoogle Storage 

Build a model from your data

Make new predictions

prediction/v1.1/training?data={}POST : a training request

prediction/v1.1/training/{}/predictGET : model infoPOST : a prediction request  

Use the API, gsutil or any compatible utility to upload your data to Google Storage

3. Predict

Page 22: Computing at scale

Example

 

Page 23: Computing at scale

Prediction API

• Google's machine learning algorithms• Available as RESTful HTTP service• Predict results in real-time

Page 24: Computing at scale

Overview

• Google App Engine• Google Storage for Developers• Prediction API• BigQuery

Page 25: Computing at scale

GET /information HTTP/1.0

• To request access and get more information, go to:o http://code.google.com/appengineo http://code.google.com/apis/bigqueryo http://code.google.com/apis/predicto http://code.google.com/apis/storage