Recovery points measured in minutes, not since last night.
Snapshot and replication design that narrows how much work you could lose — application-consistent, held off the primary system, and monitored so lag is known rather than assumed.
The gap a nightly backup leaves
A nightly backup means a failure at four in the afternoon costs a day of work. For some systems that is acceptable. For an order book, a ledger or a case management system it is not — and the gap usually goes unexamined until the day it is expensive.
- Protection is nightly, so a mid-afternoon failure loses most of a working day.
- Snapshots are enabled on the storage and nobody manages their schedule or retention.
- Replication was configured once and no one checks whether it is still current.
- Multi-volume applications are snapshotted per volume, so the copies do not agree.
What we build
Point-in-time protection tiered by how much each system can afford to lose — with consistency groups so a copy of a multi-volume application is actually usable, and lag monitored so the real recovery point is known.
- A recovery point tier per system, from minutes to daily
- Snapshot schedules and retention sized to those tiers rather than left at defaults
- Consistency groups so multi-volume and multi-database applications snapshot together
- Application-quiesced copies where the database supports it
- Replication to a second site, region or tenancy
- Replication lag monitoring, so the actual recovery point is visible rather than nominal
How it runs
Tiered deliberately — protecting everything at minute-level granularity is expensive and rarely justified.
- 01Tier by tolerable loss
Each system placed into a recovery point tier based on what losing an hour of its data would actually cost.
- 02Group for consistency
Volumes and databases belonging to one application are captured together, so the copy is coherent rather than a set of mismatched moments.
- 03Quiesce where supported
Databases are placed in a consistent state before capture where the engine allows it, which is the difference between a usable copy and a crash-consistent one.
- 04Replicate off the platform
Copies held on a separate system, site or tenancy, because a snapshot on the array that failed is not protection.
- 05Monitor the lag
Replication delay tracked and alerted, so the recovery point you rely on is the real one.
What changes once it is running
What tiered point-in-time protection changes about exposure.
The loss window narrows
Critical systems move from a day of exposure to minutes, without applying that cost to everything.
Copies are actually usable
Consistency groups and quiescing produce a restore point an application will start from.
Lag stops being invisible
The real recovery point is monitored rather than assumed from the configured schedule.
Recovery gets faster
A recent snapshot on adjacent storage restores far quicker than a full retrieval from backup media.
How an engagement is shaped
Tiering first, because it is what keeps the cost proportionate.
Tiering and capability review
One to two weeks establishing tolerable loss per system and what your existing storage, hypervisor and database platforms can already do. Often more than is being used.
Implement
Schedules, consistency groups and replication configured to the agreed tiers, with lag monitoring.
Verify and operate
Recovery points validated against the tiers, then ongoing monitoring as systems and volumes change.
Common questions
The things buyers ask before they commit. If yours is not here, it is a good first question for the assessment.
- Do snapshots replace backups?
- No. A snapshot usually lives on or near the primary system, so it does not survive the loss of that platform and is frequently deletable with the same credentials. Snapshots narrow the recovery point; backups survive the disaster. You need both.
- Does replication protect us from ransomware?
- Not on its own — it faithfully replicates the encryption, usually within minutes. Ransomware protection needs immutability and isolation, which is a different design.
- How much does minute-level protection cost?
- It scales with how much data changes and how far you replicate it. That is exactly why we tier: a handful of systems usually justify it and the rest do not.
How much work would a 4pm failure cost you?
If the honest answer is most of a day, the tiering exercise is worth doing.