September 2026 Summaries

A Tessell blog post describes a cross-cloud Oracle disaster-recovery design for an Oracle SE2 primary database on AWS and a standby on Azure, built to meet a 15-minute recovery point objective and one-hour recovery time objective without Data Guard or third-party replication software. The approach uses an IPSec site-to-site VPN between AWS and Azure, tightly scoped access for SSH and Oracle listener traffic, rsync jobs that ship Oracle archive logs every five minutes, and a continuously recovering standby database maintained in MOUNT state. It emphasizes that monitoring both archive-log apply lag and independent recovery heartbeats is necessary to detect silent replication failures, while failover requires activating the Azure standby and updating application connectivity; failback involves rebuilding AWS as a new standby. Tessell is presented as helping maintain consistent Oracle versions, patches, storage configurations, backups, point-in-time recovery, and inventory visibility across clouds, addressing configuration drift and logical-corruption risks that physical standby replication alone cannot prevent.

Tessell describes a workaround for managing large, time-based tables in Oracle Standard Edition 2, where native partitioning is unavailable and upgrading to Enterprise Edition may be too costly. For an energy and utilities customer handling continuous metering and telemetry data, the approach splits a growing table into monthly child tables and exposes them through a single UNION ALL view, preserving the existing application interface. INSTEAD OF triggers route inserts, updates, and deletes to the appropriate child table while allowing existing base-table triggers for validation, auditing, and processing to continue operating. A metadata-driven PL/SQL package, TESL_DLM_MANAGE_PARTITIONS, automates partition creation, DDL replication, constraint and index naming, privilege and synonym preservation, view rebuilding, fallback handling, retention-based cleanup, and logging. Although the framework reduces manual maintenance, supports efficient removal of old data through table drops, and avoids Enterprise Edition licensing, it does not provide optimizer-level partition pruning, its UNION ALL view expands over time, and trigger-based routing introduces additional overhead.