Upload
yan-cui
View
121
Download
3
Tags:
Embed Size (px)
DESCRIPTION
Our presentation at Google's cloud platform road show London, about our usage of BigQuery in our MMORPG Here Be Monsters.
Citation preview
BigQuery in Social GamingYan Cui, Senior DeveloperDavinder Pank, Social BI Manager
Who is Gamesys?
•Founded in 2001
•#1 in the UK
• Handle $5 Billion in turnover annually
•First company to launch real money gaming on Facebook
•Employ 1,000 globally
Travel, Collect, Craft!
Trap Monsters
Events driven Analysis
Enables deeper ad hoc analysis
Analysis goes only as far as the data
Finer the grain, bigger the volume01001001010010010010100100010101110010100010
0100100 10 1 00 10 01 0 01 01 00 10 0 01 01 01 1 10 01 01 00 0 1001001001010010010010100100010101110010100010
01001001010010010010100100010101110010100010
0100
1001
0100
1001
0010
1001
0001
0101
1100
1010
0010
0100
1001
0100
1001
0010
1001
0001
0101
1100
1010
0010
0100
1001
0100
1001
0010
1001
0001
0101
1100
1010
0010
Be Ready for Success
Jackpotjoy Slots Bingo Lane Here Be Monsters
DAU 600,000+ 150,000+ 40,000+
Events (monthly) 5 billion+ 500 million+ 700 million+
1 TB+ 200 GB+ 500 GB+
• Ensure the health of our defined KPIs across our products – Jackpotjoy Slots, Bingo Lane and Here Be Monsters
• Develop player insights to better improve the depth with which users engage with our games
Social Gaming Business Intelligence,Gamesys
BigQuery
Why BigQuery?
Scales
Managed
Fully managed architecture, allows instant project startup, and rapid time to insight
Easy to learn, minimal transition period, especially for those moving from traditional relational databases
SQL
Grows with your project, scales horizontally from 100 thousands to 100 Bn's of rows with no loss of performance on interactive queries
Monitoring KPI Health
Reporting
Apps Script Spreadsheets
Cloud Storag
e
Big Query
ETL
KPI Dashboards in Google Spreadsheets
Custom Dashboards in Google Spreadsheets
Developing Player Insights
Big JoinsNo need for temporary tables, let’s us get to the results we need in as few steps as possible
How BigQuery features and functions allow us to better explore our data
Developing Player Insights
• Our analysis hinges on being able to compare behaviour amongst players with similar in-game maturity to one another
• Where we gain some of our most invaluable insights are where changes in player behaviour lie outside the norm of what we would expect
How BigQuery features and functions allow us to better explore our data
Developing Player Insights
Window Functions
Rank and partition allow us to compare fairly player engagement and monetisation metrics across players with the same in-game maturity. Reduces dependence on external tools.
Lead and lag allow us to easily identify segments exhibiting interesting changes in behaviour
How BigQuery features and functions allow us to better explore our data
Reporting
Apps ScriptSpreadsheet
s
Cobra
App Engine
Cloud Storag
e
Big Query
ETL
Mailman
App Engine
mobile push notifications
4
facebook notifications
Beyond BigQuery
In the near future..
• Move from nightly ETL to real-time– BigQuery streaming insert
• Predication API
Thank You!
JackpotJoy Slotshttp://apps.facebook.com/jackpotjoyslots
Bingo Lanehttp://apps.facebook.com/bingolane
Here Be Monstershttp://apps.facebook.com/herebemonsters
Building a MMORPGhttp://bit.ly/1hjqoL8http://slidesha.re/18MD4XY
Google I/O 2013 – Here Be BigQueryhttp://bit.ly/1fHjbce