Technology Challenges in Building New Media Applications
ManjushaMadabushi
ManjushaMadabushi
Over 22 years of industry experience
CTO and Co-Founder, Talentica Software
Co-Founder, Nitman Software
...
Media Trends
What’s “new media”?
New media is not just digitizing of media content into bits. 
The difference really lies ...
Case Study
Ad Aggregator
171mn ads per day, 800mn+ a month
5,000+ Campaigns
150+ Carriers
5,000 Zones
Supported Ad types 
...
Traditional Ad Networks
Advertiser
(A)
Ad Network
(A)
Advertiser
(B)
Ad Network
(B)
Advertiser
(C)
Ad Network
(C)
Ad Aggregators
Advertiser
(A)
Ad Network
(A)
Aggregator
Advertiser
(B)
Ad Network
(B)
Advertiser
(D)
Advertiser
(C)
Ad Net...
Early Days
Web boxes did everything
Serving ads, run matching algorithms and measurement
Shoehorned everything into Apache...
Early Days Deployment
Load Balancer
MySQL Server
EC2
Apache Server 1
EC2
Apache Server 1
EC2
Apache Server 1
EC2
S3 Server
Challenges
Traffic increased 50% month to month
Volumes grew to 150 mn requests a day, 800mn a month
Content evolved 
Imag...
Load Balancer
Light http
Memory cache EC2
Light http
Memory cache EC2
Light http
Memory cache EC2
MySQL Server
EC2
S3 Serv...
Separate Out Priority Engine & Maintenance Jobs
Load Balancer
Light http
Memory cache EC2
Light http
Memory cache EC2
...
Scale the Database
Load Balancer
Light http
Memory cache EC2
Light http
Memory cache EC2
Light http
Memory cache EC2
MySQL...
Challenges
Traffic increased 50% month to month
Volumes grew to 150 mn requests a day, 800mn a month
Content evolved 
Imag...
Clients
Content
Clients
WAP Client
Video Ads
Iphone Client
SDK
Click to Call Ads
Android Client
SDK
Text Ad
Text Ads
Ad Ag...
Challenges
Traffic increased 50% month to month
Volumes grew to 150 mn requests a day, 800mn a month
Content evolved 
Imag...
More Reach, Less Attention
Better Targeting
Matching Algorithm to look at different attributes and return the right banner...
Targeting & Data Analysis
Hive
Hadoop Cluster
Device Location 
Time Carrier
Hive Queries
MySql Database
Reporting
Repo...
Learnings
Start Simple
Put in monitoring systems early
Use cloud for deployment
Scale quickly
Adopt component based design...
Questions, Comments…
Upcoming SlideShare
Loading in...5
×

Technology Challenges in Building New Media Applications

1,130

Published on

New media applications need to connect to millions of users across multiple media and platforms.

Applications simultaneously need to
- Work across computers, smart phones and other mobile devices, all of which have to sync up seamlessly in the background.
- Link to various social media platforms, such as blogs, twitter, facebook, etc.
- Target the content to the users of each of these platforms

This brings up its own set of technical challenges namely scalable application architectures, flexible IT infrastructure and tools for targeting

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
1,130
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Technology Challenges in Building New Media Applications

  1. 1. Technology Challenges in Building New Media Applications ManjushaMadabushi
  2. 2. ManjushaMadabushi Over 22 years of industry experience CTO and Co-Founder, Talentica Software Co-Founder, Nitman Software Director India Engineering, eGain Communications Corp Tata Consultancy Services Education MS (EECS), Northwestern University B Tech (EP), IIT Bombay
  3. 3. Media Trends What’s “new media”? New media is not just digitizing of media content into bits. The difference really lies in the dynamic life of the 'new media' content and its interactive relationship with media consumer. - Wikipedia Yearly Technology Growth 1971 1979 1991 1995 1998 2005 Internet / Web 1st Social Networking Site TV Digital Media Phones / Wireless Smart Phones Blogs / Forums Print Media Newspaper
  4. 4. Case Study Ad Aggregator 171mn ads per day, 800mn+ a month 5,000+ Campaigns 150+ Carriers 5,000 Zones Supported Ad types HTML Banner, JS Banner, SMS Banner, link banner, Image Ads, Video Ads, Iads Targeting Location, Device, Resolution, Carrier etc.
  5. 5. Traditional Ad Networks Advertiser (A) Ad Network (A) Advertiser (B) Ad Network (B) Advertiser (C) Ad Network (C)
  6. 6. Ad Aggregators Advertiser (A) Ad Network (A) Aggregator Advertiser (B) Ad Network (B) Advertiser (D) Advertiser (C) Ad Network (C)
  7. 7. Early Days Web boxes did everything Serving ads, run matching algorithms and measurement Shoehorned everything into Apache and MySQL Deployment Used Amazon EC2, S3 Monitoring Logging and monitoring using Nagios Integrations with Ad Networks Simple URLs
  8. 8. Early Days Deployment Load Balancer MySQL Server EC2 Apache Server 1 EC2 Apache Server 1 EC2 Apache Server 1 EC2 S3 Server
  9. 9. Challenges Traffic increased 50% month to month Volumes grew to 150 mn requests a day, 800mn a month Content evolved Image Ads – Standard MMA sizes / custom size Click to Call ads Video Ads User attention reduced Needed better targeting – data analysis 9
  10. 10. Load Balancer Light http Memory cache EC2 Light http Memory cache EC2 Light http Memory cache EC2 MySQL Server EC2 S3 Server 10 Scale Horizontally
  11. 11. Separate Out Priority Engine & Maintenance Jobs Load Balancer Light http Memory cache EC2 Light http Memory cache EC2 Light http Memory cache EC2 MySQL Server EC2 S3 Server Priority Engine EC2 Targetting EC2 Sumnavigation EC2
  12. 12. Scale the Database Load Balancer Light http Memory cache EC2 Light http Memory cache EC2 Light http Memory cache EC2 MySQL Master EC2 S3 Server MySQL Slave EC2 Priority Engine EC2 Targetting EC2 Sumnavigation EC2
  13. 13. Challenges Traffic increased 50% month to month Volumes grew to 150 mn requests a day, 800mn a month Content evolved Image Ads – Standard MMA sizes / custom size Click to Call ads Video Ads User attention reduced More reach – less attention
  14. 14. Clients Content Clients WAP Client Video Ads Iphone Client SDK Click to Call Ads Android Client SDK Text Ad Text Ads Ad Aggregate Products Blackberry Client SDK Image Banners Qualcomm Client SDK
  15. 15. Challenges Traffic increased 50% month to month Volumes grew to 150 mn requests a day, 800mn a month Content evolved Image Ads – Standard MMA sizes / custom size Click to Call ads Video Ads User attention reduced Needed better targeting – data analysis
  16. 16. More Reach, Less Attention Better Targeting Matching Algorithm to look at different attributes and return the right banner Location Device Device OS Resolution Carrier Time Gender, Date of Birth, Area Code Data Analysis Hadoop, Hive Clustering
  17. 17. Targeting & Data Analysis Hive Hadoop Cluster Device Location Time Carrier Hive Queries MySql Database Reporting Reports BI Reporting
  18. 18. Learnings Start Simple Put in monitoring systems early Use cloud for deployment Scale quickly Adopt component based design De-link all modules so that you can run them on separate servers Use caching Reduce load on the database servers Well defined external interfaces Connect to any provider with minimal change
  19. 19. Questions, Comments…

×