Business Intelligence Platform as a Service: Introduction to GoodData

BI Platformas a Service

Introduction to GoodData

BI PLATFORM AS A SERVICE: INTRODUCTION TO GOODDATA – CLOUDSTOCK – DEC 6, 2010 – PETR OLMER

@PetrOlmer


dashboards


report builder


data model








Data in, analytics out



ConceptWhy Business Intelligence, anyway?

GOODDATA@CLOUDSTOCK – DEC 6, 2010 – PETR OLMERBI PLATFORM AS A SERVICE: INTRODUCTION TO GOODDATA – CLOUDSTOCK – DEC 6, 2010 – PETR OLMER

BI industry is broken

too expensive

no penetration

no success


Standard IT projects

never delivered

not in time/budgetnever used

success


Standard BI projects

never delivered

not in time/budget

never used


Solution

talk cloud


Talk cloud

Be elastic.


Solution

talk simple


Talk simple


Solution

talk logic


Talk logic

It’s called business intelligence,not IT intelligence.


Talk logic

Teach a man to fish.


Talk logic

Define logical data model.

Let people play.


Data model

defines the rules of the game


Talk logic

attributes facts

attributes metrics filters

reports

dashboards


Solution

talk agile


Talk agile

Deliver one report at time.


Talk agile

Give access to everybody.


Talk agile

Don’t search for the single version of the truth.

1 cup “talk Cloud”

1 cup “talk Logic”

1 cup “talk Agile”

1 cup “talk Simple”

will feed unlimited number of people


Success recipe

1 cup “talk Cloud”

1 cup “talk Logic”

1 cup “talk Agile”

1 cup “talk Simple”

S

will feed unlimited number of people

C

L

A

S

S


Success recipe

Architecturesome clouds are more equal that others



heterogenous

multitenant

elastic

Architecture


ROLAP engine

you talk MAQL

we do SJG/QT translation

to physical level


logical layer

ROLAP engine

MAQL

TeradatamySQLIngres VectorWise PDL


Virtual cubes

be lean

think forward

decompose

reuse


Virtual cubes

many users

logical data model

MAQL

Mash Up!It’s All About Data Sources


Connectors

flat files

SQL

Salesforce

Google Analytics


Partners


Example


Data source 1


Data source 2


Connection<schema> <name>players</name> <columns> <column> <name>id</name> <title>Player Id</title> <ldmType>CONNECTION_POINT</ldmType> </column>... <schema>

<name>songs</name> <columns> <column> <name>player_id</name> <title>Player Id</title> <ldmType>REFERENCE</ldmType> <folder>Songs</folder> <schemaReference>players</schemaReference> <reference>id</reference> </column>...


Google part

google docs get --title 'Players' --format=csv --dest=.


Load dataCreateProject(name="My RockBand");

LoadJdbc(configFile="songs.xml", driver="com.mysql.jdbc.Driver",url="jdbc:mysql://localhost/sfbeta_songs", username="...",password="...", query="SELECT * FROM songs");GenerateMaql(maqlFile="songs.maql");ExecuteMaql(maqlFile="songs.maql");TransferData();

InviteUser(email="...", msg="Come and join me!", role="editor");

LoadCsv(csvDataFile="Players.csv", header="true", configFile="players.xml"); GenerateMaql(maqlFile="players.maql");ExecuteMaql(maqlFile="players.maql");TransferData();


Create metric


Create report


Create dashboard


All done

in 20 minutes

APIsWe Connect You In and Out


Browsing APIs

authentication

modeling

data loading

project provisioning


API home


If you can see it,

there’s an API call for it.


Example


Report


URI

https://secure.gooddata.com/#s=/gdc/projects/qqy0mokxwkpd7zc05mwoayz1cklyufn3¦

analysisPage¦head¦/gdc/md/qqy0mokxwkpd7zc05mwoayz1cklyufn3/obj/355

https://secure.gooddata.com



URI

https://secure.gooddata.com/#s=/gdc/projects/qqy0mokxwkpd7zc05mwoayz1cklyufn3¦

analysisPage¦head¦/gdc/md/qqy0mokxwkpd7zc05mwoayz1cklyufn3/obj/355[ ]




Report


Report definition


Data result

IntegrationA Piece Of Cake


Embedding


Integration with Chatter


Data In, Analytics Out


Step 1

load data sources

LoadGoogleAnalytics(...LoadJdbc(...


Step 2

create your metric

SELECT SUM(Visits) WHERE Location=Discussion ForumsAND Visitor=Existing Customer

GET /gdc/md/<p-hash>/obj/<metric-id>


Step 3

create report definition

filter: Week = {Last}

POST /gdc/md/<p-hash>/obj


Step 4

execute the report definition

POST /gdc/xtab2/executor


Step 5

get the value from the data result

xtab_data: columns: lookups: - 0: Visits data: - - 1620


Step 6

use Chatter API to post your message

1620 customer visits at our forums last week!


Alerts

MAQLThree Dimensions Are Not Enough


MAQL is powerful


MAQL is multidimensional


MAQL DDL

CREATE/ALTER/DROPfor model elements


Model


Model ALTER ATTRIBUTE {won}

DROP KEYS {stage},

ADD KEYS {facts_of_stages};


Model ALTER ATTRIBUTE {won}

DROP KEYS {stage},

ADD KEYS {facts_of_stages};


MAQL Queries

simple syntax for metrics

SELECT AVG(amount)

WHERE Status=Paid

BY Month, ALL Department

WITHOUT PF


Metrics in MAQL

aggregation

arithmetic

filter

dimensionality


Define metrics

Performance in Team

SELECT Team Avg Time To Response - Avg Time to Respond

Team Avg Time To Response

SELECT AVG(First Response Date - Case Created Date)

BY Team, ALL Agent WITHOUT PF

Avg Time to Respond

SELECT AVG(First Response Date - Case Created Date) BY Case


Build your report


Dimensionality

how to slice and dice


In report

What: SELECT SUM(Amount)

How: Month

... returns different number for each month


In report

What: SELECT SUM(Amount)

How: Month

Month Amount

January 1,000

February 2,000

March 3,000


Out of report

What: SELECT SUM(Amount) BY Month

How: -

... returns one numberbut different number for each month

was calculated


Out of report

What: SELECT SUM(Amount) BY Month

How: -

Month Amount

January 1,000

February 2,000

March 3,000

Amount

6,000


Out of report

What: SELECT SUM(Amount) BY ALL Month

How: Month

... returns the same numberfor each month

because only one number was computed


Out of report

What: SELECT SUM(Amount) BY ALL Month

How: Month

Month Amount

January 6,000

February 6,000

March 6,000

Amount

6,000


BY and BY ALL

BY Month, ALL OTHER

BY ALL IN ALL EXCEPT Month

WITHOUT PF


MAQL test

2 data source

Month Name Salary

January John $4,000

February John $3,900

January Paul $3,800

February Paul $3,900

January George $3,500

February George $3,500

Name Department

John IT

Paul IT

George Marketing

Ringo Marketing


MAQL test

Write a metricthat computes

average money needed monthlyfor each department.

Month Name SalaryName Department


Solution

Metrics

Monthly: SELECT SUM(Salary) BY Month

Average: SELECT AVG(Monthly)

Report

What: Average

How: Department

Data in, analytics out.



http://gooddata.com

@gooddata

http://developer.gooddata.com

@gooddata_dev

http://gooddata.com

http://gooddata.com



GoodDataIt’s BI minus the BS.

Business Intelligence Platform as a Service: Introduction to GoodData

Technology

petr olmer6

bi platformas

gooddataits bi

service introduction

architecturesome clouds

integrationa piece of

data sources

maqlthree dimensions