Snapchat's AWS Journey: Unveiling the Path of a Snap

Play video
This article is a summary of a YouTube video "Snap: Journey of a Snap on Snapchat Using AWS" by Amazon Web Services
TLDR Snapchat uses innovative architecture and services like SnapDB and Graviton to reduce latency and improve user experience.

Key insights

  • 💡
    Snapchat's features like AR lenses, maps, Bitmoji, and spotlight contribute to creating a unique and immersive user experience.
  • 💥
    Snap has over 300 million daily active users, showcasing its massive scale and popularity.
  • 💡
    Snap uses AWS services like Elastic Kubernetes Service and Cloudfront to optimize the delivery of Snaps, making them closer to the recipient when they need it.
  • 💡
    Snap chose to create their own SnapDB as a front-end to DynamoDB in order to handle higher level features like transactions, TTL, and efficient management of ephemeral data, as well as to keep costs in control.
  • 💡
    They run nightly scans on DynamoDB, processing 2 billion rows per minute, to perform various tasks such as finding friend suggestions and deleting ephemeral data.
  • 📲
    The messaging service uses Elasticache to quickly push messages to clients, ensuring efficient communication.
  • 💰
    By implementing their architecture on AWS, Snap was able to reduce the median latency of sending a Snap by 24%, resulting in a significant business impact.
  • 💡
    SnapDB is an innovative solution that reduces latency and improves user experience by cleverly handling massive amounts of data.

Timestamped Summary

  • 📷
    00:00
    Snapchat is an app that enables fast communication through features like AR lenses, maps, Bitmoji, and spotlight.
  • 💡
    00:28
    Snap has over 300 million daily active users and the first thing that happens when a user sends a Snap is the iOS app.
  • 📹
    01:06
    Snap uses an EKS service to send Snaps to Cloudfront and store them in S3, and the core orchestration service checks permissions before delivering the message to the recipient.
  • 💡
    01:44
    SnapDB was created as a front-end to DynamoDB to handle scale, higher level features, and efficient data management, storing 400 terabytes of metadata.
  • 📷
    02:30
    Snap runs nightly scans on DynamoDB to perform various tasks, including finding friend suggestions and deleting temporary data, and the receiving end of Snap is focused on minimizing latency for users.
  • 💡
    03:06
    The messaging service uses Elasticache to quickly push messages to connected clients and retrieves media and chat information through a stream using persisted media IDs.
  • 📊
    03:45
    Over 900 EKS clusters with thousands of instances each are used to handle the heavy service graph of Snapchat, resulting in a 24% reduction in median latency for sending Snaps.
  • 📝
    04:27
    Snap uses innovative architecture, such as SnapDB and Graviton, to reduce latency and lower compute costs.
Play video
This article is a summary of a YouTube video "Snap: Journey of a Snap on Snapchat Using AWS" by Amazon Web Services
4.5 (66 votes)
Report the article Report the article
Thanks for feedback Thank you for the feedback

We’ve got the additional info