RPO and RTO: what the infrastructure can actually meet
Which mechanism sets the achievable RPO and RTO, and how to read the gap to the objective.

The infrastructure meets a recovery objective through mechanisms, not intent. The achievable RPO is the distance between two recovery points — how often a backup runs, how often a snapshot is taken, whether a copy is synchronous or asynchronous. The achievable RTO is the length of the path back — how many backup sets a restore must replay, from which medium and which site. Both can be read before any failure occurs. Defining and proving the objective is governance, taken up in the sibling sheet RPO and RTO: objectives to define and then verify; this sheet stays on the infrastructure side.
Two numbers, two mechanisms
The two objectives are measured from the same instant — the failure — but in opposite directions. The RPO looks backwards: it is the gap between the last usable recovery point and the failure, fixed by the spacing of recovery points — the backup schedule, the snapshot cadence, the replication mode. The RTO looks forwards: it is the elapsed time to a service that is back in a correct operational state, fixed by the restore path — the chain that must be replayed, the medium it comes from, the site that receives it, and the check that the result is sound.
RPO runs backwards from the failure, set by the spacing of recovery points:
last recovery point failure
| |
------+--------------------------------+-----------> time
|<------- achievable RPO ------->|
| changes made after the last |
| recovery point cannot be seen |
RTO runs forwards from the failure, set by the length of the restore path:
failure service back
| |
------+--------------------------------+-----------> time
|<------ achievable RTO -------->|
| replay the chain from its |
| medium and its site, then |
| verify the result |
What fixes the RPO
The RPO is set by the interval between recovery points. NIST SP 800-34 Rev. 1 places the frequency under policy: “Policies should specify the minimum frequency and scope of backups (e.g., daily or weekly, incremental or full) based on data criticality and the frequency that new information is introduced.” Everything written between two recovery points is lost, and AWS states the arithmetic plainly: “How often you run your backup will determine your achievable recovery point (which should align to meet your RPO).” Snapshot cadence sets the same interval: an EBS snapshot “is an incremental backup”, so the exposure is the time between two snapshots.
Replication moves the interval by orders of magnitude. Microsoft’s Storage Replica documentation: synchronous replication mirrors “data within a low-latency network site with crash-consistent volumes to ensure zero data loss at the file system level during a failure”, while asynchronous replication “mirrors data across sites beyond metropolitan ranges over network links with higher latencies, but without a guarantee that both sites have identical copies of the data at the time of a potential failure”. Only the synchronous case promises that no acknowledged write is lost.
What fixes the RTO
The RTO is set by the restore path, and the chain type decides how many pieces it must assemble. NIST SP 800-34 Rev. 1 records that “to recover a system from an incremental backup, media from different backup operations may be required”, while a differential restore needs “only the full backup media and the last differential media”.
The medium and the site supply the rest. The same document measures the site’s contribution directly: “Recovery to a warm site can take several hours to several days, depending on system complexity and the amount of data to be restored”, and adds the cost rule: “the shorter the RTO, the more expensive the recovery solutions cost to implement.”
The cloud strategies are the same trade in larger steps. AWS describes four — backup and restore, pilot light, warm standby, multi-site active/active — and pilot light means servers to switch on and scale up, whereas warm standby “only requires you to scale up”. AWS advises the reader to “Use your RTO and RPO needs to help you choose between these approaches”.
Reading the gap against the objective
The objective and the achievable value are two different numbers, and the work is to put them side by side. The SNIA Dictionary keeps the terms on the side of the requirement: “Recovery Point Objective” is “The maximum acceptable time period prior to a failure or disaster during which changes to data may be lost as a consequence of recovery”, and zero “is a valid value” and is equivalent to a “zero data loss” requirement; “Recovery Time Objective” is “The maximum acceptable time period required to bring one or more applications and associated data back from an outage to a correct operational state”. Both state what is acceptable; the infrastructure supplies what is possible.
The gap closes from either end: by buying more mechanism — more frequent backups, a synchronous replica, a warmer site — or by the owner accepting a looser objective. That second is a governance decision; this sheet contributes the number the decision is made against.
Limits and the common mistake
The common mistake is to read a replication target as a recovery objective. Microsoft states the limit without hedging: “Storage Replica isn’t a backup solution.” Because the engine replicates “all changes to all blocks of data on the volume, regardless of the change type”, a deletion is copied as faithfully as a write: “If a user deletes all data from a volume, Storage Replica replicates the deletion instantly to the other volume, irrevocably removing the data from both servers.” A replica shortens the RPO for a device failure while doing almost nothing for corruption or deletion — exactly the cases where a recovery point from a backup saves the data.
The second limit is that these are capabilities, not guarantees: a recovery point never restored, on media never read back, is an assumption. Measuring that assumption against the objective is the verification the governance sheet owns.
Level and prerequisites
L3 — mechanism analysis of achievable recovery objectives: the sheet does not decide which RPO and RTO a service must have, nor design a recovery programme; it reads which storage and recovery mechanism sets each achievable value.
It stands on the L1 sheet RAID: what redundancy buys, and what it does not, and on the L2 sheet Local storage: partitions, filesystems and volume management. Among this area’s L3 sheets it depends on Full, incremental and differential backup, on Snapshots and their limits, and on Replication and the mistakes it replicates. The objective itself is the subject of the Cybersecurity sheet RPO and RTO: objectives to define and then verify.
Where to go next
To close an RPO gap, read Full, incremental and differential backup for what each chain costs to store and to restore, and Retention, immutability and separate copies for why a deletable copy is not a backup. To close an RTO gap, read Snapshots and their limits and Replication and the mistakes it replicates before selecting a site or a replication mode. To turn either value into a verified commitment, the next sheet is the governance one, RPO and RTO: objectives to define and then verify.
References
- NIST — Contingency Planning Guide for Federal Information Systems, Special Publication 800-34 Rev. 1 — https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-34r1.pdf
- SNIA — SNIA Dictionary — URL not recorded in this batch
- Microsoft — Storage Replica overview (Windows Server) — https://learn.microsoft.com/en-us/windows-server/storage/storage-replica/storage-replica-overview
- Amazon Web Services — Disaster recovery options in the cloud — https://docs.aws.amazon.com/whitepapers/latest/disaster-recovery-workloads-on-aws/disaster-recovery-options-in-the-cloud.html
- Amazon Web Services — Amazon EBS snapshots — https://docs.aws.amazon.com/ebs/latest/userguide/ebs-snapshots.html