×
Community Blog Improving Game Packaging with Kingcheer

Improving Game Packaging with Kingcheer

This article briefly introduces Kingcheer and discusses its aim to improve game packaging efficiency.

By Zhesong, Alibaba Cloud Solution Architect

Kingcheer World Shenzhen-Network Technology Co., Ltd. was established in 2016. It is an online entertainment game developer and operator integrating distribution, operation, and sales.

In the game industry, the master pack needs to be divided into different channel packages according to different promotion channels. In addition, specific channel information and channel SDK need to be integrated at the time of distribution. The size of the master pack of the game usually varies from hundreds of MB to more than 1 GB. The packaging process is constantly being carried out during game distribution every day. Due to the many game channels (more than ten channels usually), packaging between different channels is often carried out at the same time.

The process of game packaging is boring. Usually, a fixed computer is used to do the packaging according to the order of target channels, which is inefficient and troublesome. Kingcheer aims to have a simple set of concurrent processing operations to improve packaging efficiency. They also hope package machines can be released after they're done packaging, so they no longer occupy resources.

1

Serverless can the problems listed below:

1.  No Need to Worry about the Server

Managing one or two servers may not be that difficult, but managing thousands (or even tens of thousands) of servers is a different case. Faults may occur on any server. Serverless technology can automatically identify faults and remove them.

2.  Automatic Elasticity

Nowadays, Internet applications are designed to have scalable architectures. When a business has peak or trough hours or temporary traffic spikes in scenarios (such as sales promotions), the Serverless platform can smoothly scale in or scale out.

3.  Billing Based on Actual Resource Usage

Cloud resources in Serverful are billed based on resource occupancy rather than actual usage. For example, if you purchase three Elastic Compute Service (ECS) instances on the cloud, you must pay for all CPU and memory resources provided by the ECS instances regardless of how much CPU and memory you use. In Serverless models, fees are charged based on actual resource usage. For example, if you request to use the resources on an instance with 1 CPU core and 2 GB of memory for 100 ms, the money you need to pay is calculated by multiplying the unit price of the instance and the duration (namely 100 ms).

4.  Less Code, Faster Delivery

Code based on Serverless architecture usually uses backend services heavily, separating data, state management, and other contents from the code. What's more, the more thorough FaaS architecture gives the Runtime of the code to the platform management. This means the same application has much less code and supports faster distribution and startup in a Serverless architecture than in a Serverful one. Serverless platforms also offer mature features such as code establishment release and version switch to enable faster delivery.

Alibaba Cloud Function Compute (FC) is an event-driven Serverless platform that is highly elastic, cost-effective, and O&M-free. It helps developers focus on business development without procuring and managing the infrastructure (such as servers). Technological innovations of FC help enterprises quickly deliver business and realize business value.

What Are the Common Scenarios Applicable to FC?

1.  Scenarios with Sudden Traffic Spikes (Such as Flash Sales and Big Promotions)

In industries (such as new retail business, online education, and gaming), enterprises often face the problem of traffic spikes in big promotions and cannot accurately predict business peaks and troughs. At the same time, the resource budgeting process is complex, the O&M costs are high, and resource utilization cannot be guaranteed.

Hundred-millisecond level of extreme elasticity of FC solves the resource bottleneck faced by enterprises during traffic bursts. The event-driven architecture provides end-to-end solutions for more than 20 cloud products. FC provides a 1 ms granularity of pay-as-you-go billing method with 100% resource utilization, so users don't need to pay for idle resources.

2.  Live Interaction, Audio and Video Transcoding, and Other Scenarios

In such scenarios, enterprises often require intensive computing and a large number of computing resources in a short time with peaks and troughs. At the same time, the cost of self-developed services is high, and common products in the industry do not meet the highly customized requirements. Based on such pain points, FC supports a highly customized FFmpeg environment that is flexible and lightweight to ensure smooth migration without transformation.

3.  Web API/Frontend and Backend, Image Processing, and AI Scenarios

This is also one of the common scenarios of FC. Enterprises often face problems in such scenarios (such as multi-terminal adaptation, inconsistent frontend and backend standards, low R&D efficiency, and difficult O&M). In addition, data, images, and scripts are difficult to maintain. Resource utilization is low and costs are high. However, these difficulties are what FC can easily solve.

FC supports one-click Serverless deployment of web applications with multiple languages and frameworks, which significantly improves iteration efficiency. FC also supports Serverless scheduled tasks and automatic scale-out, which significantly reduces O&M costs and reduces server resource costs on a pay-as-you-go basis.

Alibaba Cloud Function Compute Helps Kingcheer Simplify Packaging, Improve Efficiency, and Reduce Costs

In order to simplify the workflow and achieve high performance and cost optimization, Kingcheer uses Alibaba Cloud Serverless Function Orchestration and Function Compute (FC) to perform packaging, which solves the pain point.

Based on the Serverless architecture, Alibaba Cloud FC provides a service platform with high availability and high computing power. It has the millisecond-level auto-scaling capability to quickly scale out and scale in the underlying layer. It can scale out the underlying layer during traffic peaks and optimize resources during traffic troughs.

2

What Benefits Does FC Bring to Kingcheer?

Efficiency Improvement: FC uses its 100-millisecond scale-out capability and task orchestration capability to process concurrent processes to improve packaging execution efficiency. The entire packaging process requires T=P* N (P is the packaging time of a single package, and N is the number of target packages) to a total execution time of P regardless of the number of channel packages, thus realizing sufficient concurrent processing.

On-Demand Cost: Compute instances are generated when an event is triggered. Resources are automatically released when computing is completed. You are only charged for on-demand instances during working hours.

Complete Monitoring: Automatic fine-grained monitoring during the execution processes (execution time and details with call status)

Customer Comments

"FC gracefully helps Kingcheer build the running environment of game packaging. The ability of on-demand concurrent execution also maximizes the efficiency of packaging. This solution also provides a direction for us to better realize the independent stateless work tasks using FC."

— Kingcheer Technical Director

0 1 0
Share on

Alibaba Cloud Serverless

99 posts | 7 followers

You may also like

Comments

Alibaba Cloud Serverless

99 posts | 7 followers

Related Products