First recorded on Twitter, sorted up:
Originally 3-4 members from the SAE team, 7-8 now
Open-source products used by Sae: Apache, ngingx, memcache, MongoDB, MySQL, etc. They are customized for various products, such as adding Apache Mod for resource allocation.
Applications deployed on SAES have multiple front-end instances,CodeFiles are synchronized through the underlying Distributed File System, and versions are controlled through memcache.
SAE implements cron jobs by means of a queue and a Distributed Global lock to ensure that the task is executed only once. The minimum heap is used to reduce polling. Other tips are not described in detail.
To avoid security issues, MySQL in SAE shields a large number of operations. Supports programming-based read/write splitting.
In Sae, the distributed database/KV-DB is not provided, which seems to be related to the next quarterly meeting. Database load or read/write splitting.
In general, Sae uses a lot of open-source software, internal infrastructure (Distributed File Systems, asynchronous queues, global locks, etc.), and large-scale systems implemented in multiple languages.
In addition, we are still very concerned about information review. Sina's colleagues said they would not delete data and the application will verify access.
Memcache plays an important role in the entire system. distributed cache can be used flexibly in many scenarios and is very useful.
In general, the SAE aggregation computing platform is built on a mature Sina infrastructure (Distributed File System, asynchronous queue, distributed lock, etc.). It uses open-source software for splicing, cropping, and customization, the entire system is finally completed. Sina's infrastructure plays a significant role in the entire system. Many core functions are implemented through these infrastructures.