1. Asynchronous decoupling
Scene description
As Taobao/TmallThe most core trading system of the main website. The generation of data on each transaction order will attract the attention of hundreds of downstream business systems, including logistics, shopping carts, points, Alibaba mother, flow calculation and analysis, etc. The overall business system is huge and complex, and the architectural design is slightly unreasonable, which will directly affect the continuity of the main website business;
High availability loose coupling architecture design
Through the loosely coupled design of upstream and downstream business systems, even if the downstream subsystem (such as logistics, points, etc.) is unavailable or even downtime, it will not affect the normal operation of the core trading system;
Flexible adaptation to rapid business growth
The mall is like a battlefield, throughMQThe asynchronous design can flexibly and efficiently adapt to changes caused by rapid business development, such as adding new business systems;
2. Peak cutting and valley filling
Scene description
Such asKill instantlyLarge-scale activities such as grabbing red envelopes and getting a good start in enterprises will bring high traffic pulses, or the system is overloaded or even crashed due to failure of corresponding protection, or the user experience is affected due to excessive restrictions. Peak cutting and valley filling is the best way to solve this problem;
Ultra-high flow pulse processing capability
MQ’s ultra-high-performance message processing capabilities can handle traffic pulses without being defeated, ensuring system availability while improving user experience due to fast and effective request responses;
Massive information accumulation capability
Ensure downstream business operates smoothly and stably within the safe water level and avoid the impact of ultra-high flow;
Reasonable cost control
By weakening the cluster size of downstream business systems, reducing input costs;
3. Sequential message
Scene description
There are many application scenarios that require order guaranteed in daily life, such as the principle of time priority in securities trading, the process of order creation, payment, refund and other processes in the trading system, and the processing of passenger boarding messages on flights. Similar to the FIFO principle, the sequential messages provided by MQ ensure the first-in-first-out of messages;
Strict order protection
Unlike some open source products, MQ always ensures the order of messages in the event of service up and down, expansion and expansion, network instability, etc.
High performance & scalable
Supports global order and partition order to meet different business needs respectively. For example, the same stocks in the securities trading system adopt global order, and different orders in the trading system adopt partition order; while strictly maintaining order, the partition order improves the overall concurrency and expansion capabilities through dynamic partition expansion capabilities;
4. Distributed transaction messages
Scene description
Alibaba's transaction system, payment red envelopes and other scenarios need to ensure the final consistency of data. The introduction of large number of distributed transactions of MQ can not only achieve decoupling between systems, but also ensure the final consistency of data;
Traditional affairs
Business processing between multiple systems or application components will be coupled to a large transaction, with long response time and long service links, which will affect the overall performance and availability of the system, and may even cause system crashes;
Distributed transactions
Split core link services with asynchronous processing branch links, split large transactions into small transactions, reduce interaction between systems, and is both efficient and reliable; MQ's reliable transmission and multi-replica technology ensure that messages are not lost, and the At-Least-Once feature ensures the final consistency of data;
5. BigData Analysis
Scene description
Data generates value in "flow", and traditional data analysis is mostly based on batch calculationsModel, but real-time data analysis cannot be achieved. Using Alibaba Cloud Message Queuing (MQ) and streaming computing engine can easily realize real-time analysis of business data.
Decoupling of application and analysis
Build a bridge between application systems and analysis systems and decouple their correlations. At the same time, because data is generated very quickly and the amount of data is large, it requires very high scalability;
Big Data Analysis
ConnectableStorm/SparkReal-time streaming computing engine, which can also be connected to offline data warehouse systems such as Hadoop/ODPS;
6. Distributed mode cache synchronization
Scene description
Tmall Double 11 promotion, a wide range of products in each sub-venue need to be perceived in real time. A large number of concurrent access databases have resulted in a long response time for the venue page. Centralized cache restricts the access traffic of product changes due to bandwidth bottlenecks. It builds a distributed cache through MQ to notify changes in product data in real time;
Real-time data update
Through real-time push of messages, data can be updated in real time;
Reduce page response time
A large number of concurrent access to the product database to reduce page response time;
Meet large-scale visit needs
The promotion is a number of sub-venues and multi-cache architecture design to meet the large number of access needs for product changes;