HTOS
About
Architecture
Scaling
Failure Modes
Fintech Notes
Tradeoffs
Design constraints change what`s best. Below are key tradeoffs and the choices made for HTOS.
Event store vs relational source-of-truth
Event store:
auditability and replay; more complexity for queries.
Relational DB:
straightforward queries and ACID; harder to rebuild state without events.
Choice:
primary relational state + event bus for replay/analytics.
Consistency vs availability
Strong consistency is simpler but reduces availability under partitions.
Eventual consistency improves availability but adds conflict handling.
Choice: consistent writes for authoritative order state; eventual consistency for derived views.
Monolith vs microservices
Monolith: easier initially; harder to scale fine-grained components.
Microservices: independent scaling at the cost of operational complexity.
Choice: modular services with clear message contracts.