AWS re:Invent 2016: Prezi's Cost-Effective, Multipetabyte Data Platform (STG307)

Опубликовано 3 декабря 2016, 3:12

Prezi has over 75 million registered users generating 900 GB of new data each day. They store XML describing over 260 million public presentations that have been viewed over 2 billion times. This created a scale problem that on-premises storage couldn't solve. In this session, you'll learn how Prezi leveraged the power and flexibility of Amazon S3 to turn their storage problem into an analytics opportunity and scale storage to meet the demands of their business without overspending. Prezi provides insight on how a small team of engineers accepted the challenge and succeeded using AWS with managed Hadoop (Amazon EMR) and Amazon Redshift. You'll see how Prezi runs and improves its data platform in a self-service data culture with this small team, without drowning in maintenance and support. Additionally, Prezi shows how the infrastructure team architected their services for storing and serving crucial data, user-generated and otherwise, using S3 and Amazon CloudFront—including pitfalls, best practices, cost considerations, and learnings along the way.