Lesson 5 — Lifecycle Processes, Change Movement, and Copies
Retention, archive, purge
- Retention: how long data must remain available because of business, legal, regulatory, contractual, or risk requirements.
- Archive: move less-active data to longer-term/cheaper storage while preserving retrieval.
- Purge: irreversibly remove obsolete/unneeded data after authorized disposition conditions are satisfied.
Old or inactive does not mean purgeable.
Capacity and growth
Capacity estimates should include more than table data: indexes, logs, redundant images/mirrors, and growth. Capacity planning is proactive; emergency expansion after exhaustion is a control failure, not a plan.
CDC vs replication vs sharding
Change Data Capture (CDC)
Detects changed data and moves deltas, using timestamps, version columns, logs, or related mechanisms.
Replication
Maintains the same data in multiple places for availability, workload distribution, geographic access, or related needs. CDC may feed replication, but the concepts are not synonymous.
- Active replication: replicas recreate/store based on operations/events from other replicas.
- Passive replication: one primary is updated; its resulting state is propagated to secondaries.
- Mirroring: secondary is kept relatively current through coordinated/immediate synchronization.
- Log shipping: transaction logs are applied to a secondary at intervals.
Sharding
Partitions one database into smaller independent chunks. Sharding divides the data; replication duplicates it.
Resiliency vs recovery
- Resiliency: tolerate errors/failure and continue functioning as expected.
- Recovery: restore service/data after loss or failure.
A resilient design may keep service running when one node fails. Recovery becomes necessary when restoration is required.
Changed-fact examples
- 1% of a huge object changed → delta/CDC/merge logic is usually more efficient than full refresh.
- nearly current failover secondary required → mirroring is stronger than periodic log shipping.
- 15-minute lag acceptable → log shipping may be sufficient.
Source: pp. 184–188, with replication/continuity applications extending into the following operational sections.