Google Cloud for Developers - Devfest Manila

Google Cloud for Developers
Apps, Apps Marketplace, App Engine for
Business, Storage, Prediction, Bigquery

Patrick Chanezon
Developer Advocate
#devfest Manila
chanezon@google.com
http://twitter.com/chanezon July 6 2010
2

Tuesday, July 6, 2010

The benefits of Cloud Computing

Economics
Pay for only what you use
TCO
OPEX vs CAPEX
Operations
Day to day: no maintenance
Fighting fires: no Pagers
Elasticity
Focus on your Business

3


Build and Buy all your enterprise cloud apps...

Google Apps Platform

Enterprise Firewall

4
Enterprise Data Authentication Enterprise Services User Management


Buy from Google

Google Apps
for Business


Enterprise Firewall

4


Buy from others Buy from Google

Google Apps Google Apps
Marketplace for Business


Enterprise Firewall

4


Buy from others Buy from Google Build your own

Google Apps Google Apps Google App Engine
Marketplace for Business for Business


Enterprise Firewall

4

Customers want more Apps

Business
in the cloud

Google
Apps

5


Google Apps Marketplace


Marketplace Overview


Google Apps is Focused on Messaging &
Collaboration

Gives every employee powerful messaging and collaboration
tools without the usual IT hassle and cost


These customers want more great cloud apps...

Messaging
Document Management
Collaboration
Productivity

Accounting & Finance Project Management

Admin Tools Sales & Marketing

Calendaring / Meetings Security & Compliance

Customer Management Workflow


Steps to sell your app to Google Apps customers
1. Build your app:
- with any tools and hosting provider you want

2. Integrate your app:
- add Single Sign On using OpenID (required)
- access over a dozen integration points from Calendar, Contacts,
Docs, etc. using OAuth (optional)

3. Sell your app:
- to 2M+ businesses & 25M+ users
- through Google Apps resellers
- to wherever Google Apps goes
- One-time fee of $100, 20% rev share starting 2H'10


Everything in the Cloud


Integrating with Google Apps & the Marketplace

Manifest


Demo Google Apps Marketplace


How it's done - Single Sign On


Single Sign On
OpenID with Google Apps

bob@foo.com Welcome Bob!
***********

Sounds complicated, but not hard in practice!


Single Sign On
OpenID Libraries
Language Libraries

Java OpenID4Java, Step2

.NET DotNetOpenAuth

php-openid,
PHP
php-openid-apps-discovery

ruby-openid,
Ruby
ruby-openid-apps-discovery

Any RPX, Ping Identity


Single Sign On - Code
Step 1 of 3 - Initialize library
function getOpenIDConsumer() {
  // Initialize client, storing associations
  // in memcache
  $store = new Auth_OpenID_MemcachedStore($MEMCACHE);
  $consumer = new Auth_OpenID_Consumer($store);

  // Enable Google Apps support
  GApps_OpenID_EnableDiscovery($consumer);
  return $consumer;
}


Step 2 of 3 - Make the request
// Create an auth request to the user's domain
$auth_request = getOpenIDConsumer()->begin($domain);

// Request email address & name during login
$ax = new Auth_OpenID_AX_FetchRequest;
$attr = Auth_OpenID_AX_AttrInfo::make(
      $AX_SCHEMA_EMAIL, 1, 1, 'email');
      $AX_SCHEMA_FIRSTNAME, 1, 1, 'ﬁrst');
      $AX_SCHEMA_LASTNAME, 1, 1, 'last');
$auth_request->addExtension($ax);

// Render Javascript/form to post request
$form_html = $auth_request->htmlMarkup($REALM,
   $RETURN_TO, false,
   array('id' => 'openid_message'));
print $form_html;


Step 3 of 3 - Handle response
// Parse the response from identity provider
$response = getOpenIDConsumer()->complete($RETURN_TO);
if ($response->status == Auth_OpenID_SUCCESS) {
  // Extract data from response
  $openid = $response->getDisplayIdentifier();
  $ax = new Auth_OpenID_AX_FetchResponse();
  $ax_resp = $ax->fromSuccessResponse($response);
  $email = $ax_resp->data[$AX_SCHEMA_EMAIL][0];
  $firstName = $ax_resp->data[$AX_SCHEMA_FIRSTNAME][0];
  $lastName = $ax_resp->data[$AX_SCHEMA_LASTTNAME][0];
  // Map to user in DB
  $user = fetch_user($openid, $email, $firstName,
   $lastName);
}


Manifest
<ApplicationManifest>
  <Name>Sassy Voice</Name>
  <Description>Voice Mail & Messaging</Description>
  <Admin>
   <Link rel="setup">
http://voice.saasyapp.com/setup.php?domain=${DOMAIN_NAME}
   </Link>
  </Admin>
  <Extension id="navlink" type="link">
   <Name>SaasyVoice</Name>
   <Url>
http://voice.saasyapp.com/login.php?domain=${DOMAIN_NAME}
   </Url>
  </Extension>
  <Extension id="realm" type="openIdRealm">
   <Url>http://voice.saasyapp.com/</Url>
  </Extension>
</ApplicationManifest>


Single Sign On
What we learned about users

Claimed ID: http://example.com/openid?id=12345

Email: bob@example.com*
First Name: Bob
Last Name: Dobbs

* Be sure to confirm or white-list trusted providers


How it's done - Data Access


2-Legged OAuth
Delegating data access

<ApplicationManifest>
  ...
  <Scope id="userFeed">
   <Url>https://apps-apis.google...
   <Reason>To get a list of user...
  </Scope>
  <Scope id="contactsFeed">
   <Url>https://www.google.com...
   <Reason>To display names of...
  </Scope>
  <Scope id="docsFeed">
   <Url>https://docs.google.com...
   <Reason>To export a call log...
  </Scope>


Data Access - Code
Step 1 of 2 - Initialize the client
function getOauthClient() {
  $options = array(
   'consumerKey' => $CONSUMER_KEY,
   'consumerSecret' => $CONSUMER_SECRET,
   'signatureMethod' => 'HMAC-SHA1',
    'requestScheme' =>Zend_Oauth::REQUEST_SCHEME_HEADER,
    'version' => '1.0');
  $consumer = new Zend_Oauth_Consumer($options);
  $token = new Zend_Oauth_Token_Access();
  $httpClient = $token->getHttpClient($options);
  return $httpClient;
}


Data Access - Code
Step 2 of 2 - Fetching Users with Provisioning API
// Initialize client
$userClient = new Zend_Gdata_Gapps(getOauthClient());
// Query feed for current user's domain
$userQuery = new Zend_Gdata_Gapps_UserQuery(
  getCurrentUserDomain());
$usersFeed = $userClient->getUserFeed($userQuery);
// Extract data from user feed
$users = array();
foreach ($usersFeed as $userEntry) {
  $login = $userEntry->getLogin();
  $name = $userEntry->getName();
  $users[] = array(
   'username' => $login->getUsername(),
   'ﬁrstName' => $name->getGivenName(),
   'lastName' => $name->getFamilyName(),
   'admin' => $login->getAdmin());
}


Integration Recap
With not a lot of code we added:

• Single Sign On with OpenID
• Quicker setup with Attribute Exchange, Provisioning API
• Integrated user data with Contacts API
• Uploading spreadsheets with Docs API


Gadgets - your real estate in Google Apps


Gadgets

• Many types of gadgets
o Gmail sidebar
o Calendar sidebar
o Sites
o Spreadsheets
• Two new types of gadgets available this week!
o Gmail contextual!
o Wave!
• All use the OpenSocial gadgets specification


Gmail contextual gadgets

• Detect e-mail content via regular expressions
• Display and collect actionable business information
directly in Gmail, below each message


Gmail contextual gadgets


Calendar sidebar gadgets
• Can detect:
o currently-displayed date range
o currently-selected event
title
attendees
date/time


Calendar sidebar gadgets


Sell


Complete Manifest

<ApplicationManifest xmlns="http://schemas.google.com/
ApplicationManifest/2009">
<Name>SaaSy Voice</Name>
  <Description>SaaSy Voice</Description>
  <Support>
   <Link rel="setup" href="https://.../setup.php?domain=${DOMAIN_NAME}"/
>
   <Link rel="manage" href="https://.../conﬁg.php?domain=$
{DOMAIN_NAME}"/>
   <Link rel="deletion-policy" href="http://.../deletion-policy.php"/>
   <Link rel="support" href="http://.../support.php"/>
  </Support>


Complete Manifest
(continued)
  
  <Extension id="oneBarLink" type="link">
   <Name>SaaSy Voice</Name>
   <Url>http://.../index.php?domain=${DOMAIN_NAME}</Url>
   <Scope ref="provisioningFeed"/>
   <Scope ref="contactsFeed"/>
   <Scope ref="docsFeed"/>
  </Extension>
  
  <Extension id="realm" type="openIdRealm">
   <Url>http://voice.saasyapp.com/marketplace/</Url>
  </Extension>

Complete Manifest
(continued)
  <Scope id="provisioningFeed">
   <Url>https://apps-apis.google.com/a/feeds/user/#readonly</Url>
   <Reason>To get a list of users to provision accounts</Reason>
  </Scope>
  <Scope id="contactsFeed">
   <Url>https://www.google.com/m8/feeds/</Url>
   <Reason>To display names of people who called</Reason>
  </Scope>
  <Scope id="docsFeed">
   <Url>https://docs.google.com/feeds/</Url>
   <Reason>To export a call log</Reason>
  </Scope>


Listing in the Google Apps Marketplace


Reach 25 million Google Apps Users

"We've seen a very meaningful increase in high quality, non-paid lead
flow directly attributable to Google Apps customers...Our customers
cite Smartsheet's tight integration with Google's Data APIs as a key
factor in their decision to purchase."
Brent Frei, Smartsheet


Roadmap


Roadmap - General
• Recent Launches
o Gmail contextual gadgets
o OAuth support for Gmail's IMAP and SMTP


Roadmap - Billing
• Goals
o Simplified invoicing & payments for customers
o Consistent experience across apps
• Planned 2nd-half of 2010
o Required for installable apps after transition period
o Revenue sharing starts once APIs adopted

• Features
o Unified billing for apps purchased in marketplace
o Support for a variety of pricing & licensing models
o APIs for customization and up-selling


Learn More


Resources
• Business and Marketing
o http://developer.googleapps.com/marketplace
• Technical and Code
o http://code.google.com/googleapps/marketplace
• Shopping!
o http://www.google.com/appsmarketplace
• Don't have Google Apps?
o http://www.google.com/a



Buy from others Buy from Google Build your own

Google Apps Google Apps Google App Engine
Marketplace for Business for Business


Enterprise Firewall

47

Google App Engine


Leveraging Google's Leadership in
Cloud Computing
• Massive data center operations
• Purpose built hardware
• Multi tenant software platform at Internet scale

49


Google App Engine

- Easy to build
- Easy to maintain
- Easy to scale

50


Cloud development in a box
• SDK & “The Cloud”
• Hardware
• Networking
• Operating system
• Application runtime
Java, Python
o

• Static file serving
• Services
• Fault tolerance
• Load balancing

51


App Engine Services

Memcache Datastore URL Fetch

Mail XMPP Task Queue

Images Blobstore User Service

52


Always free to get started

~5M pageviews/month
• 6.5 CPU hrs/day
• 1 GB storage
• 650K URL Fetch calls/day
• 2,000 recipients emailed
• 1 GB/day bandwidth
• 100,000 tasks enqueued
• 650K XMPP messages/day
53


Purchase additional resources *

* free monthly quota of ~5 million page views still in full effect
54


By the numbers

250,000+
Developers
100,000+ Apps
0.5B+ daily
Pageviews
55


App Engine

56


Socialwok

57


Social networking at scale

58


Social networking at scale

>62M Users

3.6M DAUs on Facebook
1.9M DAUs on MySpace
Orkut, Bebo, Hi5,
Friendster, Hyves, Ning, …

58


Chillingo Crystal
59
Gaming meets Social

Zombie Dash Angry Birds LITE Underground Meltdown Cogs

Mission Deep Sea Speed Forge Guerilla Bob Ravensword: Angry Birds
Extreme The Fallen King


gigy Socialize

60


gigy Socialize - traffic

61


App Engine

62


Google App Engine for Business


Build your Enterprise Apps on Google

• Easy to Build - Java standards

• Easy to Deploy - push-button deployment
• Easy to Scale - from small apps to millions of users

64


Google App Engine for Business

• Centralized administration - controls
• Reliability and support - SLA, Premium support

• Secure by default - only your users
• Pricing that makes sense - pay only for what you use

• Enterprise features - hosted SQL, SSL on your domain

65


Understanding the Cloud Computing
Landscape

SaaS

PaaS

IaaS

66
Source: Gartner AADI Summit Dec 2009


Google's Cloud Offerings

1. Our Apps
2. 3rd party Apps:
SaaS 3. ________

PaaS Google App Engine

Google Storage for Devs
IaaS Machine Learning
BigQuery

67


Google's Cloud Offerings
Your Apps
1. Our Apps
2. 3rd party Apps:
SaaS 3. ________

PaaS Google App Engine

Google Storage for Devs
IaaS Machine Learning
BigQuery

67


Domain Console

68


Domain Console

Like the regular admin console
Designed to manage enterprises with a portfolio of apps
• Keep track of all apps in a domain
• Access Control: view apps, deploy
• Global Settings: apply to all apps in the domain
• Billing rolling up to single account
• DNS configuration done only once: *.ext.example.com
• All apps by default for logged in users from domain

69


Google Apps Integration
• SSO/SSO delegation
• APIs for most Google Apps for integration

70


Federate your on-premise data

71


Secure Data Connector (SDC)

72


Using Secure Data Connector

73



Installation
- Determine access rules
- Configure and install SDC

73



Installation
Getting ready to serve
- SDC opens SSL tunnel

73



Installation
Getting ready to serve
- SDC opens SSL tunnel
Serving
- User request sent to App Engine
- User authenticated
- App makes request through tunnel
- SDC performs access checks
- Results returned
73


App Engine for Business Pricing
Intranet apps:
Each app costs $8 / active user / month
Capped at $1,000 / month (i.e. users above 125 are free)
Apps are auth-restricted to domain users
Development is free
Overage charges on Background Analysis/Storage

Non intranet apps (external/public/ISV apps):
Pricing TBD

Postpaid (i.e. billed at the end of month)

74


App Engine for Business Support and
SLA
Paid Support
Email based
1000$/month
1h response time on operational issues
8h on development issues

SLA
99.9% uptime
Service credits from 10% to 100% refund of monthly bill

75


Google Developer Qualification

Chrome Gadgets Search App Engine JS Maps API KML 3D
Extensions

76


Two years in review
Apr 2008 Python launch
May 2008 Memcache, Images API
Jul 2008 Logs export
Aug 2008 Batch write/delete
Oct 2008 HTTPS support
Dec 2008 Status dashboard, quota details
Feb 2009 Billing, larger files
Apr 2009 Java launch, DB import, cron support, SDC
May 2009 Key-only queries
Jun 2009 Task queues
Aug 2009 Kindless queries
Sep 2009 XMPP
Oct 2009 Incoming email
Dec 2009 Blobstore
Feb 2010 Datastore cursors, Appstats
Mar 2010 Read policies, IPv6
May 2010 App Engine for Business
77


An evolving platform

78

App Engine
Roadmap

Improved monitoring/alerting
Background servers
SSL for your domain
Control datastore availability vs. latency trade-offs
Datastore dump and restore facility
Mapping operations across datasets
Raise request/response size limits for some APIs
Reserved instances
Built-in support for OAuth & OpenID
Channel API 79


Google Storage


Google Storage
• Cloud-based binary object store
o Structured as buckets and objects
o Many buckets, many objects, large objects

• You control your data
o Private, shared, or public
o Get your data back out at any time

• For developers
o RESTful API
o Many SDKs + tools
o Integration with other Google services


Google Storage Benefits

High Performance and Scalability
backed by Google infrastructure

Flexible Authentication &
Sharing Models

Get Started Fast with
Google & 3rd Party Utilities


Demo: Get Started In < 1 Minute
• click emailed invitation link

• read & accept Terms Of Service

• start using GS Manager


Google services using Google Storage

Partner Reporting

Data Liberation Haiti Relief Imagery Partner Reporting

Google Google
BigQuery Prediction API


Some current users


Technical Overview


Google Storage Overview

• Fast, scalable, highly available object store
o Objects of any type and practically any size
o Lots of objects, lots of buckets
o All data replicated to multiple US data centers
o Read-your-writes data consistency

• Easy, flexible authentication and sharing
o Key-based authentication
o Authenticated downloads from a web browser
o Sharing with individuals and groups

• Google products and 3rd party tools/services
o Compatible with many available tools and libraries
o Getting started toolkit


API Concepts

• RESTful API
o Verbs: GET, PUT, POST, HEAD, DELETE
o Resources: identified by URI

• Buckets
o Flat containers

• Objects
o Any type, practically any size

• Access Control for Google Accounts
o Google Groups

• Two Ways to Authenticate Requests
o Sign request using access keys
o Web browser login


Sample Signed Request

PUT /mybucket/My/Long/Object/Name HTTP/1.1
Host: commondatastorage.googleapis.com:443
Accept-Encoding: identity
Date: Sat, 08 May 2010 19:04:21 GMT
Content-Length: 28
Content-Type: text/plain
Authorization:GOOG1
GOOG4622809698762217:J+y3mj5GThfI6Ed1MqLi7JpCq5Y=

This is my object's content.


Sharing and ACLs

• Data can be private or shared

• Bucket ACL determines:
o who can list objects (READ)
o who can create / delete objects (WRITE)
o who can read / write bucket ACL (FULL_CONTROL)

• Object ACL determines:
o who can read objects (READ)
o who can read / write object ACL (FULL_CONTROL)


Read-Your-Writes Consistency

• Once a write succeeds, all future reads will see a snapshot
that includes that write... no matter what replica they talk to

• Once any reader sees a result (even if the write previously
appeared to fail) then all future readers will see a snapshot
that includes the write


Interoperability Gives You Choice

• Data Liberation
o You shouldn't be locked in by your choice of
storage provider

• Choice of tools
o You should be able to use the same tools to
manage your data, regardless of where you keep it


Demo: Using Google Storage


Computing in the Cloud: App Engine Sample
os.environ['BOTO_CONFIG'] = 'boto.cfg'
from boto import storage_uri
# <other imports omitted>

class MainPage(webapp.RequestHandler):
  def get(self):
   self.response.out.write('<html><body>')
   uri = storage_uri('gs://pub/shakespeare/rose.txt')
   poem = uri.get_contents_as_string()
   self.response.out.write('<pre>' + poem + '</pre>')
   self.response.out.write('</body></html>')

def main():
  application = webapp.WSGIApplication([('/', MainPage)])
  run_wsgi_app(application)

if __name__ == "__main__":
  main()


Pricing and Availability

• Pay as you go pricing
• Storage - $0.17/GB/month
• Network
o Upload data to Google
 $0.10/GB
o Download data from Google
 $0.15/GB for Americas and EMEA
 $0.30/GB for APAC
• Requests
o PUT, POST, LIST - $0.01 per 1,000 Requests
o GET, HEAD - $0.01 per 10,000 Requests

• Free storage (up to 100GB) during preview period
o No SLA
• http://code.google.com/apis/storage


What's Coming Up

• Service Level Agreement

• Support

• Available to Premium Apps Customers

• Technical Features:
o Group support in ACLs
o Resumable uploads
o Additional regions


Google BigQuery and Prediction APIs


Overview

• Big Data - Challenging and Important

• Google has tools for deep data analysis

• Now you can use these tools


Overview

• Big Data - Challenging and Important

• Google has tools for deep data analysis

• Now you can use these tools

• Announcing two new APIs to get more from your data:
1.BigQuery
2.Prediction API


Benefits

• Built on Google technology

• Scalability

• Security

• Sharing

• Easy integration with Google App Engine, Google
Spreadsheets, ....


Using Your Data with BigQuery & Prediction API

1. Upload Upload your data Your Data
to Google Storage

Import to tables BigQuery
2. Process Train a model Prediction API

Run queries Your Apps
3. Act Make predictions


BigQuery
Interactive Analysis of Big Data


Big Data is Challenging
Starts with Scale


Many Use Cases ...

Interactive Tools Trends
Spam Detection

Web Network
Dashboards Optimization


Demo: Analyzing M-Lab
An open platform for advanced network research
http://www.measurementlab.net/


Demo: Exploring M-Lab


Key Capabilities of BigQuery

• Scalable: Billions of rows
• Fast: Response in seconds
• Simple: Queries in SQL
• Web Service
o REST
o JSON-RPC
o Google App Scripts


Using Your Data with BigQuery

Upload to Google Storage
1. Upload

Import data into a BigQuery Table
2. Import - No need to define indices, keys, etc..

Execute queries via APIs
3. Query - No provisioning machines or resources


Writing Queries

Compact subset of SQL
o SELECT ... FROM ...
WHERE ...
GROUP BY ... ORDER BY ...
LIMIT ...;

Common functions
o Math, String, Time, ...

Statistical approximations
o TOP
o COUNT DISTINCT


API in a Minute
GET /bigquery/v1/tables/{table name}

GET /bigquery/v1/query?q={query}
Sample JSON Reply:
{
  "results": {
   "ﬁelds": { [
   {"id":"COUNT(*)","type":"uint64"}, ... ]
   },
   "rows": [
   {"f":[{"v":"2949"}, ...]},
   {"f":[{"v":"5387"}, ...]}, ... ]
  }
}

Also supports JSON-RPC


Security and Privacy

Standard Google Authentication
• Client Login
• OAuth
• AuthSub

HTTPS support
• protects your credentials
• protects your data

Use Google Storage for Developers to manage access


Large Corpus Analysis
Wikimedia Revision History

Wikimedia Revision history data from: http://download.wikimedia.org/enwiki/latest/
enwiki-latest-pages-meta-history.xml.7z


Using BigQuery Shell
Python DB API 2.0 + B. Clapper's sqlcmd
http://www.clapper.org/software/python/sqlcmd/


BigQuery from a Spreadsheet


BigQuery Recap

• Interactive analysis of very large data sets
• Simple SQL query language

• APIs enable a variety of use cases


Google Prediction API
Machine learning as a web service


Prediction API 101

• Google's sophisticated machine learning algorithms

• Available as an on-demand RESTful HTTP web service

• Train a model offline/asynchronously

• Predict results in real-time

"Tous pour un,
un pour tous, Prediction API "french"
c'est notre
devise."


How does it work?

"english" The quick brown fox jumped over the
The Prediction API lazy dog.
finds relevant
"english" To err is human, but to really foul things
features in the up you need a computer.
sample data during
"spanish" No hay mal que por bien no venga.
training.
"spanish" La tercera es la vencida.

The Prediction API
later searches for ? To be or not to be, that is the question.
those features
during prediction. ? La fe mueve montañas.


A virtually endless number of applications...

Customer Transaction Species Message Diagnostics
Sentiment Risk Identification Routing

Churn Legal Docket Suspicious Work Roster Inappropriate
Prediction Classification Activity Assignment Content

Recommend Political Uplift Email Career
Products Bias Marketing Filtering Counselling

... and many more ...


Three simple steps to use the Prediction API

Upload your Use the API, gsutil or any
1. Upload training data to compatible utility to upload
Google Storage your data to Google Storage

Build a model prediction/v1/train/{}
2. Train from your data POST : a training request

prediction/v1/query/{}
Make new GET : model info
3. Predict predictions POST : a prediction request


Prediction API Demo
Automatically categorize and respond to emails by language

• Customer: ACME Corp, a multinational organization
• Goal: Respond to customer emails in their language
• Data: Many emails, tagged with their languages

• Outcome: Predict language and respond accordingly


Step 1: Upload
Upload your training data to Google Storage
• Training data: outputs and input features
• Data format: comma separated value format (CSV)

$ head -n 2 ${data}
"english","To err is human, but to really ..."
"spanish","No hay mal que por bien no venga."

Upload to Google Storage
$ gsutil cp ${data} gs://io10/${data}


Step 2: Train
Create a new model by training on data

To train a model:

POST prediction/v1/train/${data}

Training runs asynchronously. To see if it has finished:
GET prediction/v1/query/${data}

{"data": {
   "resource": {
   "data": "${data}",
   "modelinfo": "estimated accuracy: ${acc}"}}}


Step 3: Predict
Apply the trained model to make predictions on new data

POST prediction/v1/query/${data}
{ data : {
  "instance" : {
   "input" : { "text" : [
   "J'aime X! C'est le meilleur" ]}}}}


Step 3: Predict

POST prediction/v1/query/${data}
{ data : {
  "instance" : {
   "input" : { "text" : [
   "J'aime X! C'est le meilleur" ]}
   "output" :
   {"output_label" : "french"}}}}


Step 3: Predict

import httplib

header = {"Content-Type" : "application/json"}#...put new data in
JSON format in params variable
conn =
httplib.HTTPConnection("www.googleapis.com")conn.request("P
OST",
"/prediction/v1/query/${data}", params, header)print
conn.getresponse()


Prediction API Capabilities
Data
• Input Features: numeric or unstructured text
• Output: up to 100s of discrete categories

Training
• Many machine learning techniques
• Automatically selected
• Performed asynchronously

Access from many platforms:
• Web app from Google App Engine
• Apps Script (e.g. from Google Spreadsheet)
• Desktop app


Prediction API and BigQuery Demo: Tagger
Input Data: http://delic.io.us/chanezon
– 6000 urls, 14000 tags in 6 years
Analyze my delicious tags
– use delicious API to get all tagged urls
– cleanup data, resize (100Mb limit)
– PUT data in Google storage
– Define table
– analyze
Predict how I would tag a technology article
– input is tag,url,text
– send new url and text
– get predicted tag


Get the BigQuery & Prediction APIs

• Preview, opened to a limited number of developers
• You need a Google Storage for Developers account
• To request access and get more information, go to:
o http://code.google.com/apis/bigquery
o http://code.google.com/apis/prediction


Acknowledgement

Thanks to many Googlers from their slides, most
available at the Google IO 2010 website

131


Thank you

Read more
http://code.google.com/appengine/

Contact info
Patrick Chanezon
Developer Advocate
chanezon@google.com
http://twitter.com/chanezon

Questions
?

132


Google Cloud for Developers - Devfest Manila

More Related Content

What's hot

Viewers also liked

Similar to Google Cloud for Developers - Devfest Manila

More from Patrick Chanezon

Recently uploaded

Google Cloud for Developers - Devfest Manila