AWS-based advertising service application architecture
The Internet advertisement service needs to complete targeted advertisements within a limited time, which is a technical challenge. AWS provides reliable, fault-tolerant, and highly available system infrastructure and services built on the cloud. This article describes two main parts of the system: the basic structure of the Advertising Service and the data analysis cluster used for clicking data collection.
1. When a user loads a web page, the ad server returns the ID displayed for an ad resource. The ad server runs on an EC2 instance and queries the relevant ad information stored in the Amazon DynamoDB table based on your personal data.
2. Download ad files from Amazon CloudFront. This content publishing service has no submitted content and has low latency and high data transmission rate. The log information of the displayed advertisement exists on S3.
3. The click Collection Server is a group of Amazon EC2 instances used to collect user click data. This information is stored in the log file of the click collection server and periodically uploaded to Amazon S3.
4. Amazon Elastic MapReduce is a parallel job stream that uses the Hadoop framework to process data. Ad effect and click data are collected and processed by the Amazon Elastic MapReduce cluster. By using hotspot instances, the cluster capacity can be dynamically expanded while reducing the processing time and the cost of running the job flow.
5. Data processing results are pushed back to Amazon DynamoDB, a fully manageable NoSQL database service that provides fast and predictable performance and seamless scalability. Amazon DynamoDB can store and obtain a large amount of data, serve different levels of traffic requests, and quickly store and obtain users' personal data. Whether it's a massive user data set or a high-traffic request, Amazon DynamoDB's high availability and robustness can ensure that the front-end of the ad server responds in a short time.
This article permanently updates the link address: