Ad Serving Platform
Architecture for the real-time bidding advertising platform
About Architecture
Real-time bidding (RTB) is a technology to purchase and sell the advertisement inventory displayed on the web or the mobile screen through a program type instant auction method. It is a real-time auction system where all processes take place individually and instantly so that the advertisers can display advertisements to the optimal user at an optimal price, and the publishers can sell the advertisement inventory at the best price. Such bidding processes must be handled within tens of millisecond. All processes including the network connection and analysis must be handled without latency because the advertisement URL must be returned by analyzing the advertisement display size of the requested website and the past data of the user to analyze the bidding price. The task of making valuable information by analyzing all traffic of the user is also important, and must be made to be utilized in deciding the bidding price. Tasks and the result of profiling by calculating the efficiency of specific advertisement for the user and tracking/recording the user's behavior using the technology, such as machine learning, must be easily stored and managed. The NAVER CLOUD PLATFORM is effective since you can build an analysis system that can be linked using various methods such as the server instance, which enables scaling by default and the repository to read/write quickly to make all these processes possible.
Architecture
7-1_AdServingPlatform_1601268959355.png)
Related Services
Use Cases and Effect
- Building a Flexible Server Instance
- The Prixcell server for analyzing the bidding requests and user logs that come in through the real-time bidding system must enable fast scaling to prepare for instantly increasing traffic or an unpredictable spark. You can prepare for the required computing resources flexibly through the products like Load Balancer and Auto Scaling.
- Collecting and Handling the Traffic for Analysis
- To ensure high availability, two master nodes are provided for redundancy when creating a Cloud Hadoop cluster, the managed type product. In addition, you can create an automated advertising system by collecting and analyzing the user traffic since it is possible to use by linking with various open sources. You can receive and handle stream data through Kafka, utilize as a real-time analysis system using the machine-learning feature of Spark ML, and build a promotion plan by analyzing the data stored in Object Storage or RDB for a long period of time.
- Operation of an Economical RTB Platform
- One of the key advantages of cloud is that effective cost management is possible. The infrastructure of the NAVER CLOUD PLATFORM is designed for optimal cost effectiveness for your business and you can manage your operational costs by paying based on your usage amount.
- Availability and Convenience of Object Storage
- Save large amount of data at a low cost using Object Storage on the NAVER CLOUD PLATFORM as a data storage service. You can use without worrying about the capacity since easy extension is possible at a reasonable cost from a gigabyte unit to a petabyte unit according to the customer's business scale, and you can also link to analyze data in Cloud Hadoop.