OK S3

Odnoklassniki team has been developing and maintaining distributed fault-tolerant custom storages for many years. Our hot/cold blob storages provides efficient handling and processing of binary data (video, photo, music, etc.) approaching 1EB in size. Cassandra-based NewSQL supports our transactional workloads in numerous high-performance distributed clusters. Apart from custom homemade systems we have been using a fleet of deployed on-premises external products to store and deliver build and test artifacts, libraries, images, packages, logs, backups, etc. External Amazon S3 or on-premises S3 compatible storage service may fit the role of the common storage backend for all those products. S3 provides a high level object API and an evolved ecosystem of tools and SDKs for the most popular languages and platforms. In his talk Vadim will describe implementation of S3 compatible storage service based on battle-tested blob and NewSQL storages at Odnoklassniki. The talk will cover the architecture and data model, features and trade-offs, performance and optimization, some intricacies and surprises. The ideas and experience presented in the talk might be useful to designers and developers of distributed storage services, especially S3-compatible and/or Cassandra-based ones.