Same ~36 MB binary ships every deployment pattern. What varies
by topology is the node count, the replication model, and the
RTO/RPO commitment. The engine is identical across all four.
PICK A TOPOLOGY·LIVE DIAGRAM + SLO TABLE
NODES
1
RTO
Single point — restore from snapshot
RPO
WAL fsync (sub-second)
AVAILABILITY
Best effort
USE WHEN
POC · dev · CI · edge site · air-gapped lab · single-tenant small corpus
NODES
3 · one region
RTO
< 30 s · automatic replica promotion
RPO
0 (synchronous replication)
AVAILABILITY
99.95 %
USE WHEN
Standard production · typical enterprise footprint · 1–5 B docs
NODES
3 + 3 · two regions
RTO
< 2 min · DNS failover
RPO
Sub-second (async)
AVAILABILITY
99.99 %
USE WHEN
Global enterprise · > 10 B docs · region-of-choice compliance · disaster recovery scope
NODES
1 · loopback by default
RTO / RPO
Operator-defined
EGRESS
No runtime telemetry · update · license calls; Console may attempt external Google Fonts
UPDATES
Operator-staged archive · adjacent .sha256 integrity only
USE WHEN
Classified enclaves · SCIFs · regulated on-prem with no internet · sovereign cloud