This topic uses JL-Tour as an example and describes the integration between JL-Tour and Function Compute, Message Queue for Apache RocketMQ, and Object Storage Service (OSS). The example in this topic shows you how to process instantaneous concurrent messages by using these Alibaba Cloud services.

Customer introduction

Established in 2001, Shenzhen JL-Tour International Travel Service Co.,Ltd. (JL-Tour) has been focusing on the development of hotel distribution services and the information construction for hotel booking. The company has built a leading hotel distribution system that targets the B2B business model. JL-Tour keeps true to their philosophy of offering convenient, cost-effective, and efficient services, and is committed to building the best B2B tourism ecosystem in China. JL-Tour already forges strong cooperation ties with more than 10,000 tourism distribution channels inside and outside China, and is offering resources from more than 600,000 starred hotels in more than 300 cities globally.

JL-Tour has developed deep cooperation relationships with leading online travel agencies (OTAs) for years. LY.COM, Qunar, Tuniu.com, Expedia, Ctrip, Agoda, Lvmama, Mafengwo, CITS American Express Global Business Travel (CITS GBT), and Datang Business Travel are all on the customer list of JL-Tour for hotel booking and ticketing services.

Customer pain points and requirements

JL-Tour receives more than ten million pieces of data updates from more than 600,000 hotels every day. These data updates require high concurrency and feature short validity period. This poses great pressure on the existing system of JL-Tour in terms of instantaneous concurrent message processing. JL-Tour needed an advanced system that can provide the following features and reduce the pressure of concurrent message processing:
  • Concurrent processing capacity: The system must concurrently process a maximum of 100,000 messages.
  • Scalable processing capacity: The system must automatically scale in or out within milliseconds based on the number of messages to be processed. The cost of use is charged based on the actual resources required.
  • Support for multiple data sources, such as OSS and messages.
  • Support for multiple programming languages, such as Python, Go, and Java.
  • O&M monitoring capabilities: The system allows for rapid deployment and updates, monitors real-time resource usage, analyzes logs, and generates alerts.

Solution

By taking the following advantages of Function Compute and by using Message Queue for Apache RocketMQ and OSS, JL-Tour meets its business requirements:
  • Function Compute listens to a variety of data sources.
    • It monitors and processes changes in the volume of business, and carries out adaptive scale-out and scale-in operations with efficiency.
    • It monitors scale-out operations that are performed within milliseconds. This allowed JL-Tour to achieve linear growth in business capacity.
  • Function Compute supports multiple programming languages for easy use.
  • Function Compute supports easy deployment and can monitor real-time resource usage, analyze logs, and generate alerts.

Benefits of Function Compute

  • Business stability: Function Compute automatically scales in or out based on the volume of business. By using Function Compute, JL-Tour can ensure business stability without allocating resources based on spikes in resource usage.
  • Simple O&M: Function Compute allows JL-Tour to improve O&M capability by using a variety of tools. Function Compute eliminates the need of scalable resource management and also enables the O&M engineers of JL-Tour to work with higher efficiency.
  • Cost efficiency: Function Compute adopts the pay-as-you-go billing method. JL-Tour can select proper specifications to ensure that its resources are efficiently utilized, without the need to pay for idle resources. The overall usage cost of JL-Tour is reduced.