Background of the project

In 2020, 37InteractiveEntertainment has achieved rapid growth in overseas markets. In order to maintain continued business growth in 2021 and expand the overseas influence of the mobile game business, the Sanqi Mobile Game Division plans to build a global mobile game platform and integrate relevant platform resources independently built according to individual game needs in the early stage to accelerate new games The release speed of the game and the unified collection of game user data.

37InteractiveEntertainment currently adopts the PL-ES service model. Agilewing helps customers use cloud vendors services very well. In the past year, Agilewing, cloud vendors and 37InteractiveEntertainment have established a close cooperative relationship and assisted 37InteractiveEntertainment online games division to complete the architecture optimization. Has fully won the trust of customers, so the customer decided to deploy this global mobile game platform on cloud vendors.

Currently, all workloads are deployed in On Center, and applications and databases are deployed through self-built methods. The server has been running for a long time, the server failure rate is high, and the maintenance cost is high. At the same time, for future development, it is hoped that this migration can easily transform applications and use managed services as much as possible.

Target

• 1.Migrate the workloads currently running on IDC to cloud vendors, and provide best-practice deployment architecture to achieve rapid access by users worldwide
• 2.Complete the transformation of application hosting services, and provide feasible monitoring operation and maintenance solutions
• 3. Complete the construction of the data analysis platform. The core cloud vendors services are Elastic MapReduce(EMR) and Redshift.

Solution

Effect

Agilewing helps customers complete the design, construction and key process testing of the entire platform to achieve rapid platform deployment and launch. At the same time, using Elastic MapReduce (EMR) service to reduce the work pressure of the operation and maintenance team and the cost of the data analysis platform

1. 1. DMS data synchronization dimension table data to Redshift
2. 2. Sqoop exports aurora data to S3, hive creates an external table to query the data exported by Sqoop
3. 3. Hive builds internal tables to store DW layer data
4. 4. Aurora enables binlog to be synchronized to MSK (kafka) via DMS
5. 5. Copy S3 data into redshift
6. 6. Create redshift resource queue
7. 7. Redshift merges the temporary table data to the fact table (using the merge queue)
8. 8. Redshift uses queues for queries
9. 9. Redshift Federated Query Aurora (Preview)