Glide.me Leverages Redis Cloud to Scale Their 200G In-Memory Database

技术

分析与建模 - 实时分析
应用基础设施与中间件 - 数据交换与集成
基础设施即服务 (IaaS) - 云存储服务

适用行业

Software
电信

适用功能

商业运营
质量保证

用例

预测性维护
实时定位系统 (RTLS)
远程资产管理

服务

云规划/设计/实施服务
数据科学服务
系统集成

挑战

Glide faced multiple challenges in scaling their real-time messaging platform. One major challenge was the need for real-time reminders and analytics. Every time a user received a message, a reminder had to be created, requiring push notifications in real-time. This necessitated persistent in-memory storage across servers and real-time aggregated analytics to support admin performance reports. Another challenge was a client-side bug that caused session tokens to be corrupted, leading to new login requests for every interaction. This resulted in a 'memory leak' in Redis, causing unnecessary data volume growth. Lastly, Glide faced issues with Pubnub message counters due to the large volume of messages created daily, which exceeded expected limits.

关于客户

Glide is the world's first mobile video instant messenger, allowing users to send text and video messages via live stream. The platform offers the flexibility to read or watch messages in real-time or later. Hosted on AWS and Redis Cloud, Glide leverages a massive real-time network to deliver consistent performance for millions of users. The application is available on Android and iOS operating systems and uses a combination of AWS services and Redis Cloud for backend hosting and data storage. Glide's innovative approach to mobile messaging has made it a popular choice for users seeking real-time communication.

解决方案

To address the challenges, Glide implemented several solutions leveraging Redis Cloud. For real-time reminders and analytics, Redis Labs' platform enabled Glide to store and share unread and missed messages efficiently. Aggregated analytics were incremented to a counter in Redis, which later persisted to a backend database for admin panel viewing. This approach minimized strain on the backend database. To resolve the session token bug, Glide moved sessions to a new database to isolate the issue and then addressed the faulty application logic. Redis Cloud professionals guided the process, ensuring minimal impact on the application's service. For Pubnub message counters, Glide used Redis to increment a counter every time a message was sent, facilitating easier tracking for Pubnub metrics. Redis Cloud's auto-scaling capabilities ensured seamless performance despite increased workload.

运营影响

Redis Cloud enabled Glide to store and share unread and missed messages efficiently, reducing the need for repeated database queries.
Aggregated analytics were incremented to a counter in Redis, minimizing strain on the backend database and facilitating real-time performance reports.
The session token bug was resolved by isolating the issue in a new database and addressing the faulty application logic, drastically reducing session storage space.

数量效益

Glide's Redis database can now scale up to 400GB, providing ample room for future growth.
The session storage issue was resolved, significantly reducing the amount of space needed for session storage.