当前位置:网站首页>Best practices for elastic computing in the game industry
Best practices for elastic computing in the game industry
2022-07-29 05:27:00 【Elastic computing Bai Xiaosheng】
Search for the wild , Alibaba cloud elastic computing product solution architect
Abstract : Games have always been a hot industry in the Internet field , With the rise of mobile Internet , Mobile phones and Pad The popularity of , The game started in the early days PC Page tour of the times 、 The swim , Gradually developed to mobile games occupy the main trend . In recent years , Game manufacturers are constantly exploring cloud games , Hope to bring a smoother experience to players , The cloud has become the best choice for the game industry .
How to make the game run more stably on the cloud 、 Fluent 、 Efficient and economical ? As the largest cloud manufacturer in China , Alibaba cloud provides game manufacturers with the best cloud solution . This article mainly introduces three game scenes : Game services 、 Big data operations 、 The architectural features of cloud games , And the selection and deployment scheme of Alibaba cloud game industry computing infrastructure based on these scenarios .
One 、 The recalculation scenario of the game industry
The game industry is a scene of recalculation power consumption , Three of the scenarios where the demand for computing power is more prominent are : Game clothes 、 Game operation 、 Cloud games .
l Game clothes
There are many kinds of games , This article will focus on two types of computing intensive in game clothing :MMORPG The type and MOBA/FPS type .
l Game operation
Several major trends in the current game industry include : Category integration and playing methods are constantly superimposed 、 Social class DAU The trend of this kind of game 、 And the trend of boutique games , This requires a lot of data operation analysis , And help game companies improve user retention 、 Optimize the playing methods and strategies in the game .
l Cloud games
This article will introduce the practice of elastic computing products in combination with the above three scenarios .
Two 、 Game service business and technical features
1. Game service business features
From the game category level , Yes RPG、FPS、MOBA、SLG Other categories , Each category has its own characteristics in terms of the playing method of the game itself .
From the perspective of technical architecture , Due to the strong interactivity of the game , Game technology architecture is different from other Internet applications :
a. A long connection
Games usually require long connections , That is, a long connection from a client to a server , Facilitate timely feedback and push of player's behavior ;
b. Low latency
Due to the strong interactivity of the game , No matter in PVE still PVP in , All need timely feedback , similar FPS、MOBA Class and other multiplayer games are typical PVP, The tolerance of network delay and jitter is very low , So low latency is required ;
c. Write frequently
Due to the high frequency of player information recording in the game 、 The characteristics of the operation , These records are recorded when there are a large number of players, such as typical MMO game , Need to write data frequently , Concurrent reading and writing is very high , This kind of scene needs strong IO performance ;
d. High performance
The game is highly interactive 、 Features of low latency , In today's gradually layered decoupling Architecture , We also need to ensure the interaction effect of game players , Will further rely on the strong computing power of the underlying server ;
e. High elasticity
Cycle characteristics and peak valley characteristics of game business , Very high elasticity is often required .
2. MMORPG Elastic computing product practice
RPG( Role playing games ) It's a big category of games , Players simulate a virtual character to experience and play in the game .
a. MMORPG The characteristics of game clothes
- MMO: Massive players, high concurrency scenes , High computing performance 、 Strong stability ;
- Gateway Service : Network packet forwarding 、 High throughput , High network PPS;
- Center Service : Minute level asynchronous warehousing , Write frequently 、 high IOPS;
b. Elastic computing product practice
Based on the above characteristics ,MMORPG The selection of elastic computing products should also be distinguished according to the scene :
- MMO: High computing performance , Seven generations Ice Lake ECS example , Improve the performance experience ;
- Gateway Service : High network PPS, Suggest 6 Generation enhancement 、7 Generation examples ;
- Asynchronous warehousing and logging : For scenarios with high disk read-write performance , Use different levels ESSD Cloud disk , Avoid disk read / write bottlenecks ;
- Efficient operation and maintenance : For partitioned Services , Scenes such as opening and closing services quickly , It can be done by CADT Yunsuda 、ESS Stretch and stretch 、OOS Choreography and other tools (3) It can be done by CADT Yunsuda 、 Stretch and stretch 、 The use of cloud operation and maintenance tools such as operation and maintenance arrangement with products can improve the operation and maintenance efficiency .
3. FPS、MOBA Elastic computing product practice
image FPS/MOBA This kind of competitive game , More interactive , Delay is more sensitive .
a. Features of war games
- Low latency : A lot of computing intensive scenes in the game ;
- Ultra high performance : Big map battle 、 Small room battle ;
- Elastic room : Rolling switch in combat room .
b. Elastic computing product practice
- Deployment form : Platform function center , Combat clothing area deployment ;
- Low latency and high performance : High dominant frequency 7 generation 、7 Generation computing instance , Higher single core performance provides better combat effectiveness ;
- Elastic room : Battle room games , Due to the peak and valley characteristics of the business itself , Flexible use of cloud resources , It will often better optimize the overall resource use cost ,ESS Stretch and stretch + The cloud monitoring CMS+ Saving plan , Reach the cost optimal solution ;
- Containerization :ACK Container services +ECI Examples of flexible containers , It further releases the flexibility and elasticity of basic resources ;
- Platform business : This part is closer to the general Internet technology architecture , For example, the wonderful real-time recording scene is a typical high-definition image quality + Demand for real-time recording , Often based on GPU Capacity building , Selective elasticity vGPU/cCPU product , Achieve finer grained use of... On the cloud GPU resources .
3、 ... and 、 Game big data elastic Computing Practice
Business performance in the whole life cycle of the game , If retained by the user 、 Business transformation 、 In game playing methods and strategies are the business support data that game manufacturers pay close attention to .
Big data is the current game business 、 Operation and other main technical means , Thanks to the richness of the big data open source community technology stack , There are many technical options for big data , The optional scheme shall be adjusted to local conditions according to its own project .
1. Game big data business features
- Data scenarios : Business operation data 、 Game content data 、 Platform technical data 、 Safe operation data ;
- Technology stack selection : Technology selection 、 Deployment options 、 Operation and maintenance pressure ;
- Efficiency issues : Business performance 、 Cost burden 、 Elastic demand .
2. Elastic computing practice
a. Based on cloud products
Different game companies are in different stages of data construction , There will be different choices , Included in IDC build by oneself 、 Self built based on cloud based resources 、 On the cloud EMR Suite hosting 、 And use more big data products on the cloud , Improve business efficiency and reduce the burden of operation and maintenance , such as : choice SLS Data collection ;Kafka Data gateway channel ; adopt Flink Do real-time calculation ; adopt Elasticsearch or Clickhouse Do data aggregation analysis ;
b. Selection idea releases efficiency
The specific product selection of computing storage also needs to be based on business requirements 、 performance 、 efficiency 、 Balance costs in many ways . Such as real-time calculation / Near real-time computing scenario ,Flink High performance 、 Low latency features , Computing and network performance are required , Give priority to the selection of seven generation main sales examples or 6 Generation enhancement instance ; And for Remote Shuffle Service Scenes with many intermediate processing results , A lot of join Class scenes , You can choose to partially process through memory ; Offline data calculation and storage may choose SATA Local disk instance ; as for SSD Local dish 、SATA Local dish 、ESSD Cloud disk , We need to take a comprehensive look at the calculation 、 The Internet 、 Choose from storage performance and capacity as well as comprehensive cost ;
c. Release elasticity
be based on OSS The data Lake scheme of object storage is also one of the trends , The advantage of the decoupling architecture of storage and computing on the cloud is that it fully releases the ability of elasticity on the cloud , You can also use more lightweight elastic computing products , Such as SPOT example ,ECI Calculate elastic big data for container instances , So as to meet the business needs and save more costs ;
d. Safety operation and maintenance
Account security management , Local dish D2S Heat exchanger plate , And event triggering mechanism .
Four 、 Cloud game elastic computing product practice
from 2009 year ONLIVE Launch cloud game products and start , Cloud games have passed a technology cycle , Especially in the last two years , More and more companies pay attention to and invest in cloud games . The platform is based on fluidization capability , Drive the game in the form of video streaming , Enable users to enjoy higher quality games at low cost .
Cloud games are mainly divided into terminal and cloud . The terminal part is based on Windows、IOS、Linux Terminal equipment of operating system , Including mobile phone 、 Flat 、 The computer 、 The TV 、VR All-in-one, etc. .
The cloud architecture is mainly the game application layer 、 Cloud game platform layer 、 Basic resource layer . The application layer is a huge amount of game content ; The platform layer is the necessary operation platform for cloud games 、 Support platform 、 Fluidization technology platform, etc ;IaaS The resource layer includes the basic network 、 be based on X86 Architecture and ARM Architecturally GPU The server .
1. Cloud game business features
- Multi terminal adaptation : Mobile devices 、 Home appliances 、 Scene equipment ;
- End to end experience : Achieve end-to-end high performance and low delay , Instruction stream 、 Edit code ;
- Graphic calculation : HD game rendering
2. Elastic computing product practice
- Enrich GPU example : High performance GPU(3A)、vGPU( Page to swim )、ARM+GPU( Cloud hand tour );
- Performance acceleration experience : Streaming technology 、 Rendering acceleration 、 Hardware acceleration ;
- center + Edge scene coverage : Public cloud + Edge cloud box ;
- Flexible construction : Rapid deployment 、 Elastic use ;
- Yunyunwei experience :CloudOps、 Cloud speed CADT etc. , Help game customers to build cloud game business more quickly and conveniently .
5、 ... and 、 Alibaba cloud provides high-quality game infrastructure services for global game developers
Through years of technology accumulation and continuous operation , At present, it provides very large-scale infrastructure cloud services all over the world , Through excellent and stable performance , And the cost dividend continuously released through technical means , Provide game customers with more benefits 、 More secure 、 A more stable 、 More flexible infrastructure cloud services .
Domestic business goes to sea 、 Game going to sea is also one of the main trends in the domestic game market , Many game companies have turned going to sea from a business option to one of the necessary options .
2022 year 3 month , Alibaba cloud has also launched two regions, South Korea and Thailand , It can provide a smoother and more stable game experience for the localization service of the game . Alibaba cloud will be in the business field of game going to sea , Provide more help to customers .
Click on here , Check the wonderful sharing videos of guests
边栏推荐
- 如视技术副总裁杨永林:当传统产业遇到“数字空间”
- R & D efficiency | analysis of kubernetes' core technology and Devops' landing experience
- CMU15-213 Shell Lab实验记录
- CryEngine3 调试Shader方法
- 365天挑战LeetCode1000题——Day 038 公交站间的距离 + 基于时间的键值存储 + 转变数组后最接近目标值的数组和 + 有界数组中指定下标处的最大值
- Best practices of JD cloud Distributed Link Tracking in financial scenarios
- OCCT学习003-----MFC单文档工程
- C语言文件操作
- 什么是_GLIBCXX_VISIBILITY(default)
- CMU15-213 Malloc Lab实验记录
猜你喜欢
哈夫曼树以及哈夫曼编码在文件压缩上的应用
JD cloud and Forrester consulting released a hybrid cloud report that cloud Nativity has become a new engine driving industrial development
MySQL的基础概念+数据库系统结构+拓展延申+基础命令学习
QT series - Installation
C语言数组入门到精通(数组精讲)
[event preview] cloud development, efficient and intelligent - the second Alibaba cloud ECS cloudbuild developer competition is about to start
OCCT学习001-----简介
510000 prize pool invites you to fight! The second Alibaba cloud ECS cloudbuild developer competition is coming
如视技术副总裁杨永林:当传统产业遇到“数字空间”
365 day challenge leetcode 1000 questions - day 037 elements and the maximum side length of squares less than or equal to the threshold + the number of subsequences that meet the conditions
随机推荐
时间复杂度和空间复杂度
365 day challenge leetcode1000 question - distance between bus stops on day 038 + time-based key value storage + array closest to the target value after transforming the array and + maximum value at t
如视技术副总裁杨永林:当传统产业遇到“数字空间”
QtCreator+CMake编译器设置
365天挑战LeetCode1000题——Day 035 每日一题 + 二分查找 13
递归的基本原理
阿里云联合鼎捷软件发布云上数字工厂解决方案,实现云MES系统本地化部署
最新坦克大战2022-全程开发笔记-3
365 day challenge leetcode 1000 questions - day 039 full binary tree inserter + find peak II + snapshot array
CMake 设置vs启动运行环境路径
京东云分布式链路追踪在金融场景的最佳实践
321,京东言犀×NLPCC 2022挑战赛开赛!
Qt版的贪食蛇游戏项目
水一篇图的拓扑排序
存储类别
D3d Shader Instruction
小白高薪捷径-Qt开发游戏—贪吃蛇
C语言用指向指针的指针对n个整数排序
终端shell常用命令
ANSI C类型限定符