forked from diegoecab/oci-deal-accelerator
initial version
This commit is contained in:
71
kb/patterns/database-ha-adb-s.yaml
Normal file
71
kb/patterns/database-ha-adb-s.yaml
Normal file
@@ -0,0 +1,71 @@
|
||||
# Pattern: ADB-S High Availability
|
||||
# Composable architecture block for Autonomous Database Serverless HA
|
||||
|
||||
pattern:
|
||||
name: "ADB-S High Availability"
|
||||
id: database_ha_adb_s
|
||||
category: database-ha
|
||||
|
||||
description: |
|
||||
Autonomous Database Serverless (ADB-S) provides built-in high availability through
|
||||
Transparent Application Continuity (TAC) and automated failover. TAC replays in-flight
|
||||
transactions transparently after infrastructure failures, masking outages from applications.
|
||||
For additional protection, Autonomous Data Guard provides a same-region standby with
|
||||
synchronous replication for zero data loss.
|
||||
|
||||
variants:
|
||||
- id: local_ha
|
||||
name: "Built-in Local HA (TAC)"
|
||||
topology: "Same-region, ADB-S infrastructure-managed failover"
|
||||
rpo: "0"
|
||||
rto: "~0 perceived (transparent replay via TAC)"
|
||||
use_when:
|
||||
- "Standard HA for ADB-S workloads"
|
||||
- "Application can tolerate brief transparent replay"
|
||||
- "No separate standby database required"
|
||||
|
||||
- id: local_standby
|
||||
name: "Same-Region Autonomous Data Guard Standby"
|
||||
topology: "Primary + Standby in same region via Autonomous Data Guard"
|
||||
rpo: "0 (synchronous replication)"
|
||||
rto: "< 30 seconds (automatic switchover)"
|
||||
use_when:
|
||||
- "Zero RPO mandatory"
|
||||
- "Protection against full AD or FD failure"
|
||||
- "Read-only offload to standby needed"
|
||||
|
||||
preconditions:
|
||||
- "ADB-S instance provisioned (Shared infrastructure)"
|
||||
- "Private endpoint recommended for production workloads"
|
||||
- "VCN with Service Gateway for secure OCI service access"
|
||||
- "Wallet or mTLS connectivity configured for clients"
|
||||
|
||||
implied_services:
|
||||
- service: service_gateway
|
||||
reason: "Secure access to OCI services without internet traversal"
|
||||
- service: oci_vault
|
||||
reason: "Customer-managed encryption keys for TDE"
|
||||
- service: data_safe
|
||||
reason: "Database security posture management, auditing, and masking"
|
||||
|
||||
gotchas:
|
||||
- "TAC replay does not cover all operations — UTL_HTTP calls, DBMS_PIPE, NOCACHE sequences, and external interactions are not replayed"
|
||||
- "Auto-scaling takes 2-3 minutes to add OCPUs; plan base capacity for burst tolerance"
|
||||
- "Autonomous Data Guard standby is read-only; cannot run DML on standby"
|
||||
- "Switchover requires brief downtime; failover is automatic on primary failure"
|
||||
- "TAC requires JDBC Thin driver 18c+ or ODP.NET with TAC enabled in connection string"
|
||||
|
||||
cost_impact:
|
||||
- "Built-in local HA (TAC): no additional cost, included in ADB-S"
|
||||
- "Autonomous Data Guard standby: billed at same compute (OCPU/ECPU) rate as primary"
|
||||
- "Standby storage is additional and billed at standard ADB-S storage rates"
|
||||
- "Auto-scaling can increase OCPU cost up to 3x base during peaks"
|
||||
|
||||
sizing_rules:
|
||||
- "Standby must be same shape and OCPU/ECPU count as primary for seamless failover"
|
||||
- "Storage on standby mirrors primary automatically"
|
||||
- "Base OCPU count should handle normal workload without relying on auto-scaling for HA events"
|
||||
|
||||
references:
|
||||
- "https://docs.oracle.com/en/cloud/paas/autonomous-database/adbsa/autonomous-data-guard.html"
|
||||
- "https://docs.oracle.com/en/cloud/paas/autonomous-database/serverless/adbsb/application-continuity-autonomous.html"
|
||||
Reference in New Issue
Block a user