A Smart Bear · Jason Cohen
articleprincipleDistributed Logical Time
23 February 2019Profit
Source excerpt
About A Smart BearA simple, decentralized, scalable, constant-memory mechanism for independent replicas to record events in time, preserving the "happened-before" relation in almost all cases.
This is a limited feed-provided excerpt, not the full original work.
Apollo layer
What a founder can learn
Founder takeaway
When independent replicas must order events, favor mechanisms that preserve causal relationships without central coordination while keeping memory and scaling costs bounded. Treat “almost all cases” as an explicit product tradeoff, not an implementation footnote.
Why it matters
Distributed architectures can improve scalability and resilience, but ambiguous event ordering may create incorrect states or behavior. Founders should ensure the chosen consistency model matches the consequences of getting rare cases wrong.
Relevant guides
Put it to work
Ask your technical team: Which workflows require exact causal ordering, which can tolerate rare ambiguity, and how will the system detect or resolve exceptions?