Game Cloud Game Distributed advanced Deployment Architecture Scheme

Distributed Deployment architecture: implementation of the application of loose coupling, will Web application server and caching, database services and other modules decoupling, and physically isolated deployment, so that the level of functional modules can be extended, but also to avoid the cloud server IO scramble to bring about the performance degradation problem.

To eliminate a single point of deployment:

Add more than one login server, the back-end game area deployed multiple units, are providing the same service externally; both are configured load balanced sharing flow, to avoid single point of failure.

Game Update Service:

Can be based on Aliyun OSS+CDN provide static file distribution of low latency, if the game has a large number of pictures, audio and video, maps and static data distribution, you can also consider through the OSS+CDN to ensure the game experience.

Using the RDS server:

is a stable and reliable, flexible and scalable online database services. RDS is the use of out-of-the-way, compatible with MySQL, SQL Server two relational database, and provide database master-slave hot standby, online expansion, backup rollback, performance monitoring and analysis functions, the current maximum performance of RDS 12000IOPS.

To enable caching servers:

As the user volume increases, the pressure on the backend DB access will increase, and we can increase the cache server (OCS or REDIS) storage hotspot data between game server and DB, such as the leaderboard, to increase the response speed of the access and reduce the access pressure on the backend db.

Massive Log data analysis:

Aliyun provides a massive data analysis service similar to Hadoop (ODPs), users only need to import the game log or dump database files into the system, you can use most of the SQL statements to do data analysis of the amount, ODPs support large table jion, stored procedures and a variety of statistical algorithms , the data mining and analysis based on ODPs can be done, and the method of cluster provides service, which greatly enhances the analysis time of massive data and greatly facilitates the use of massive data by users.

Game Distributed advanced Deployment architecture diagram

The game Distributed Advanced Deployment Architecture Interpretation:

The
1. load Balancing
2. SLB) extends the service capabilities of the application system through traffic distribution, and eliminates the single point of failure to elevate the availability of the application system. The SLB bottom is a cluster mechanism, with throughput and stability far higher than the load-balancing hardware and software that is deployed, and requires only lower cost inputs. On
3. logon server
4. ECS, the load Balancing service is enabled to extend the number of servers in real time, sharing user traffic, using ECS's resilient extension capabilities and load-balanced streaming capabilities as traffic increases. For example, chess games will be held regularly, the number of online users increased, for games requiring player identification, it is recommended that the login server use high-end configuration, and enable load balancing (SLB) services.
5. Game Server
6.  
7. security Protection
8. GB-level anti-DDoS cleaning service that allows users to turn on the 7-layer cleaning service to protect against CC attacks. Can also selectively open the free intrusion prevention and web site application Security monitoring services.
9. 10 types of real-time monitoring services, such as CPU, memory, and network traffic, and provide a variety of alerting methods.
10. Game Database
11. , SQL Server database scenario: A game with relational databases that can be built on your own cloud server or deployed to relational database services (RDS). RDS is a stable and reliable, flexible and scalable online database service, compatible with MySQL, SQL Server two relational databases, and provides online expansion of the database, backup rollback, performance monitoring and analysis functions, with the use of RDS and cloud server to improve I/O performance.
12. Database Scenario: You can build MongoDB on a cloud server, because the MongoDB run is large in memory, it is recommended to purchase X64 system, and select the cloud server with large memory configuration according to the actual demand.
13. Cache server
14.  
15. caching scheme: an open cache service (OCS) that supports KEY-VALUE data structures, and clients that are compatible memcached protocols can communicate with OCS. Use OCS in combination with RDS to form a high throughput, low-latency storage solution. Placing hotspot data in OCS can greatly ease the load on the database, not only to save costs, but also to improve system availability.
16. Caching scheme: It can be realized by building Redis on ECS. Redis Applicable TourMore scenes, such as: Game player charts, using the ordered set provided by Redis, support each key value (such as player ID) to have a score (Score), each time to add elements to this set, Redis will be sorted, modify the Score of an element, will also update the sort, When fetching data, you can specify the sort range. More importantly, this sort result is saved without being recalculated at server startup. The list data type provided by Redis can be used to implement a message queue. Because it is independent of the game server, multiple game servers can be used to exchange data, send events.
17. content Distribution
18. CDN high-quality edge nodes throughout the country, scheduling precision, with enterprise-class DDoS defense capabilities, easy to use. In the game, a large number of high pixel pictures, video and other static resources in the transmission will often lead to poor customer experience, Aliyun CDN Technology through the establishment of a wide range of data nodes, static resources are cached in the nearest node, so as to speed up the transfer of resources and improve the user experience.
19. file storage
20. OSS, which stores static and frequently accessed files such as game clients, upgrades, and so on, can reduce server bandwidth pressure and effectively respond to high concurrency. Its chargeback mechanism can reduce financial risk and cost.
21. Data Analysis
22. Hadoop makes large data analysis, but the use and maintenance of the Hadoop cluster is even more complex than the data analysis logic itself, and it is based on this consideration that Aliyun's large data Processing service (ODPS) By abstracting the data analysis cluster platform as a service, you need to focus on the data analysis logic itself without caring for the deployment and maintenance of the cluster.
23. >>