Space-Based Architecture
Scale and resilience via partitioned, replicated in-memory data grids and collocated processing; minimizes database contention.
What it is
A style that distributes state and processing across an in-memory grid. Requests are routed to partitions holding relevant data; replicas provide failover.
Benefits
- Very low latency and high throughput
- Resilience via replication and partition isolation
- Reduces load on centralized databases
Trade-offs
- Complexity in partitioning and data affinity
- State synchronization and consistency challenges
- Operational overhead for grid management
When to use
- High-throughput, low-latency domains (trading, ecommerce)
- Hot data sets amenable to partitioning
- When DB contention or latency is a bottleneck