Loyal Guru 利用 AlloyDB 将查询延迟降低 40-50% | 谷歌云博客_AI阅读总结 — 包阅AI

Editor’s note: Loyal Guru, a loyalty and offer personalization platform, uses scalable and secure architecture from Google Cloud to help its retail customers deliver highly personalized experiences in real time. By leveraging Google Cloud databases like AlloyDB for PostgreSQL, the company overcame performance bottlenecks and reduced query latency by 40-50%.

At Loyal Guru, we understand the challenges retailers face with engaging customers and driving loyalty amidst rapidly evolving expectations. Our loyalty and offer personalization platform allows retail clients to analyze customer data and tailor promotions dynamically to match precise preferences and context. The goal? Helping retailers create long-lasting relationships through greater relevance.

For us, innovation is vital — the retail world won’t stop evolving, so neither can we. We knew a robust platform would help deliver truly personalized experiences quickly and at scale. That’s why we partnered with Google Cloud for our infrastructure and databases. AlloyDB, along with other Google Cloud products like Firestore and BigQuery, provides the agility and performance needed to continually enhance our platform’s capabilities and help us anticipate emerging trends rather than merely reacting.

Powering personalization with high-performance databases

Loyal Guru is designed to help retailers deeply engage customers through tailored promotions, offers, and brand experiences. Delivering this at scale, however, requires robust data processing and analytics capabilities. Retailers need to ingest and act on customers’ behavioral data in real time across multiple touchpoints to create precise audience segmentation and dynamic offer adjustments that maximize relevance.

Powering these mission-critical personalization functions demands a high-performance, scalable database. Our customers need instant access to customer data, offer details, transactional records, and more for querying and updates as customer interactions occur. Overloaded databases lead to slow responses and disappointing customer experiences. That’s why we chose to build our platform on top of Google Cloud databases.

A secure, scalable database strategy

At the core of our architecture are three powerful Google Cloud data products: AlloyDB for PostgreSQL, Firestore, and BigQuery.

AlloyDB for transactional workloads
We knew that Google Cloud databases offer exceptional scalability and performance, but we didn’t expect the superior performance and fast analytical capabilities we get from AlloyDB, which serves as our relational database. We rely on AlloyDB to power all of our core transactional processes, including user profiles, loyalty program details, and offer redemptions. It ensures our APIs have reliable, low-latency access to this critical data. It also provides a strong foundation to start using the power of AlloyDB and Gemini to build generative AI apps that can help expand our offerings.

Prior to adopting AlloyDB, large traffic spikes from major retail events on Loyal Guru often created bottlenecks. We struggled to deal with data growth, which strained the limits of our storage capacity. AlloyDB’s autopilot feature for managing storage has been instrumental in mitigating these issues, automatically adjusting storage capacity to meet our demands without manual intervention. Overall, we reduced query latency by 40-50% and optimized storage from 2TB to 1.3TB — a 35% improvement.

Firestore for high-concurrency reads
We use Firestore in Datastore Mode for high-throughput read operations on relatively static data, such as loyalty points balances and product catalogs. Its ability to handle massive concurrency without compromising speed makes it ideal for these read-heavy workloads. This allowed our platform to maintain high availability and performance even under the heaviest read loads from retailers using our services.

BigQuery for analytics
To extract maximum value from our data assets, we process large behavioral datasets through BigQuery for analytics and business intelligence. As our data warehouse solution, we use BigQuery to run complex queries, visualize trends, and surface insights that guide strategic decisions. These analytics capabilities are critical for optimizing our personalization models, campaign strategies, and product roadmap based on our customers’ data. BigQuery’s scalable and fully managed infrastructure also enables us to handle massive datasets and complex queries at high speeds without getting bogged down with infrastructure limitations..

In addition to these three data products, we also leverage multiple Google Cloud components to help us provide a secure service with high availability. The Google App Engine flexible environment allows us to automatically scale our Loyal Guru APIs, and we host our progressive web app (PWA) for managing campaigns with Firebase Hosting. We also utilize Cloud Load Balancing together with Cloud Armor for traffic distribution and cloud-native security. With Cloud Balancing we’re able to improve scalability and reliability of our services by improving rotation and escalation policies, at the same time we filter invalid/attack attempts before arriving to the GAE Service. This in conjunction with Cloud Armor, which give us the ability to detect som attack attempts like XSS, SQLi, and other common malicious practices.