Case Study

December 8, 2017

Poing – High Availability Solution for Advertising Platform


Case Studies



Poing Interactive is a US invested start-up company, focusing on building a new way of delivering ads into games. Their main products are In Game Advertising Platform, and Playscreen.

The In Game Advertisement Platform is a platform that meant to delivers fully integrated ads into games. Using this platform, an ad can be fully embedded into a game, as its assets. Players will engage with ads as they engage with other game’s assets or properties.

Playscreen is a new breed of arcade. It’s an Android PC with a 42 inch touchscreen TV, now placed in Shopping Malls, where people can be introduced, play and download registered games for free.

1. The challenges

There are 4 systems running businesses of Poing: Advertisements, Reward, Management and CDN systems on DigitalOcean infrastructure.

  • Each system contained a single linux server for webserver and an installed MariaDB inside those servers, which was too risky.
  • No Virtual Private Cloud (VPC), which was difficult to manage subnets and security
  • Using HA proxy server for Load balancing, which can be overload and limited on numbers of served requests and bandwidth
  • Using Linux server for CDN system, which was not fully CDN. It was only a webserver that returned static files.
  • Unable to scale up when system must serve too much requests from end-users.

2. Our solution

The architecture we applied for Poing includes:

  • VPC: to manage Private Network likes traditional VPN
  • MariaDB: Using RDS MariaDB – Multi AZ
  • Webservers: Using EC2s for Java Tomcat
  • HA proxy: Using ELB instead of HA proxy
  • Auto Scale setup
  • Storage: S3, using SDK for Java
  • Hazelcast: autodiscovery feature to auto dicover EC2 member instances.
  • CDN: replace original CDN with CloudFront Distribution

3. The outcome

  • Database: no failover problems by using RDS Multi-AZ and automatically backup snapshots daily
  • No record any system outage status.
  • Servers: 2 instances running in normal traffic, scale up to 4 or 5 instances when increases spikes
  • Scalability & High availabilty: setup succesfully high availibity and scalability system on AWS infrastructure
  • Cost optimization: reduce infrastructure cost by 30%
  • ELB: work properly, serving unlimited requests, never die endpoint
  • Storage: no data loss incident reported so far
  • CDN: work properly

Copyright © 2018 - OSAM. All Rights Reserved.