Under the Hood of a Shard-per-Core Database Architecture

Explore the inner workings of a shard-per-core database architecture in this 47-minute webinar sponsored by ScyllaDB. Delve into the advancements of modern server technology and examine ScyllaDB's innovative approach, including its asynchronous engine, the Seastar framework. Learn how this architecture avoids CPU-level locks and contention, bypasses kernel bottlenecks, implements a per-core shared-nothing autosharding mechanism, and utilizes modern storage hardware. Discover how NUMA is leveraged for optimal RAM performance and how data is balanced across CPUs and nodes for smooth, efficient operation. Gain insights into the advantages of unlocking vertical scalability and understand how this architecture addresses performance issues, reduces overprovisioning, and lowers total cost of ownership compared to traditional database architectures.