Tair (Enterprise Edition) offers a variety of data models and enterprise-grade features, enabling Shihuo to develop a robust online business system that effortlessly supports large-scale promotions and manages the challenges of millions of QPS with ease.
Customer testimonials
"Tair (Enterprise Edition) enables the Shihuo app to maintain a system architecture that supports high concurrency and low latency, adeptly managing frequent access to hot data and effortlessly handling traffic spikes of tens of millions of users during major e-commerce events such as Double 11 and the 618 Shopping Festival. This ensures that users receive up-to-date and engaging discounts on sports, trends, lifestyle, fashion, and more." — Shihuo Operations Team Owner
Customer profile
Shihuo is a discount product guide platform committed to offering professional online shopping decision support. It delivers up-to-date and exciting online shopping discount information on sports, trends, lifestyle, fashion, and more, catering to cost-conscious users. The platform features products from major shopping malls both in and outside China.
Business challenges
Strong Demand for Ultra-High Concurrency
As Shihuo's business rapidly expands, user numbers have skyrocketed, and the system has scaled quickly. To consistently deliver a stable and superior user experience, Shihuo requires a platform capable of supporting high concurrency and low latency.
Challenges of Traffic Surges During E-commerce Flash Sales
Flash sales, a popular marketing strategy on e-commerce platforms, often lead to traffic spikes during events like Double 11 and the 618 Shopping Festival. Ensuring a seamless user experience amid intense requests for hot products is a significant challenge for system stability.
Exponential Growth in Data Volume and Rising Cost Concerns
With increasing industry competition, the cost of acquiring new customers has risen. Shihuo focuses on analyzing existing users to enhance service. Leveraging behavioral data, Shihuo performs profile analysis and implements personalized precision marketing solutions, which necessitates the storage and analysis of vast data volumes, highlighting cost concerns.
Stability Risks Posed by Large Keys and Hotkeys
Using open-source Redis, large keys and hotkeys can cause issues due to inadequate business planning, accumulation of stale data, or sudden traffic increases. These issues can lead to service jitter, performance degradation, poor user experience, and even widespread outages. Prompt identification and optimization solutions for large keys and hotkeys are essential.
Solutions
Tair as a Cache Layer Supports High-Concurrency, Low-Latency System Architecture
Tair, as a high-speed caching technology, can significantly enhance system performance and availability. Tair offers a cloud-native architecture that is immediately operational and meets business demands for high throughput, low latency, and dynamic scalability. Its cluster architecture allows for the flexible adjustment of shard numbers ranging from 2 to 256 and supports the modification of shard specifications, easily handling tens of millions of concurrent access requests.
Proxy Query Cache Technology Supports Hot Data Access Requests
In e-commerce flash sales, hot-selling products naturally become hot data. Tair detects hotkeys within instances using efficient sorting and statistical algorithms. The proxy node caches both requests and query results for hotkeys based on predefined rules. If the same request arrives within the cache's validity period, the proxy node immediately returns the result to the client, eliminating the need to connect with the backend data shard for each request and reducing network transmission overhead. This enhances read speeds, mitigates the impact of hotkeys on data shard performance, and prevents request skewness. Thus, Tair supports Shihuo in managing high-concurrency, read-heavy scenarios with frequent access to hot data.
Tair Persistent Memory Aids in Efficiency and Cost Reduction
Alibaba Cloud has introduced a proprietary engine designed for data persistence, leveraging persistent memory and integrating it with Shenlong bare metal servers to launch the Tair persistent memory product. Shihuo opted for Tair persistent storage instances to manage the extensive data from their business systems. The performance rivals that of open source Redis, yet the overall costs are reduced by 30%, striking an optimal balance between performance and cost, and effectively enhancing efficiency and cost savings.
Tair Facilitates Easy Identification of Large Keys and Hotkeys
Tair offers a wide range of dimensions, various types, and enhanced observability capabilities. It combines performance metrics, slow query logs, and key analysis to deliver comprehensive, one-stop instance diagnosis functions. By evaluating the health of instances from multiple perspectives, including performance levels, access distribution, and slow query logs, it provides actionable improvement recommendations. This significantly boosts the automated operation and maintenance of Tair instances. This reduces costs and assists customers in quickly identifying and resolving system stability issues.
Customer value
Tair, serving as a cache layer, supports tens of millions of high concurrency accesses, enabling customers to deliver a stable and superior user experience.
With Tair's assistance, Shihuo successfully manages traffic spikes during significant e-commerce promotions like Double 11 and the 618 Shopping Festival.
Tair has reduced the cost of persistent memory by 30%, enabling customers to achieve greater efficiency and cost savings.
Tair offers comprehensive solutions to address challenges encountered in usage, significantly reducing operational and maintenance costs.
Tair memory instances offer triple the read and write performance of comparable native Redis or open source Redis solutions, enabling businesses to surmount performance bottlenecks associated with frequent access to hot data.
