ISC · Information Systems and Data Management · 9 practice questions
Choose Disaster Recovery Architecture by RTO and RPO
A disaster recovery design must meet both the recovery-time limit and the data-loss limit. Below: compare a synchronous active-active design with alternatives that change only replication or recovery time.
Try one first
Hint
Focus on which option both prevents any committed transaction from being lost at the secondary site and supports immediate, automated takeover without manual rebuild or extended restores.
Answer C. Synchronous replication ensures each committed transaction is mirrored to the secondary before the primary acknowledges it, guaranteeing RPO = 0. Automated failover or automatic load‑balancing allows the alternate site to take over without manual restoration, enabling an RTO within the 15‑minute window. Options relying on asynchronous replication, log restores, or nightly backups risk data loss or exceed the short recovery time.
Why not A: Tempting because continuous log shipping is simpler and improves on nightly backups. But restoring systems and replaying logs takes time and may not capture the very latest commits at failure; log shipping without synchronous commit does not guarantee RPO = 0 and the restore process is likely to exceed a 15‑minute RTO.
Why not B: Tempting because a warm standby is cheaper and frequent asynchronous replication reduces exposure. However, asynchronous replication can lag and allow committed transactions to be lost at failover (violates RPO = 0), and scripted/manual failover usually produces an RTO longer than 15 minutes.
Why not D: Tempting because local clustering handles hardware faults quickly and is familiar to many teams. However, a cold site rebuilt from nightly backups allows up to a day of data loss (violates RPO = 0) and will take far longer than 15 minutes to recover, failing both SLA requirements.
Side by side
| Base: synchronous active-active | Same facts, except asynchronous replication | Same facts, except 12-hour recovery | |
|---|---|---|---|
| Required outage protection | QuickShip: complete data center loss | QuickShip: complete data center loss | QuickShip: complete data center loss |
| Business recovery limits | RTO: 5 minutes; RPO: 1 minute | RTO: 5 minutes; RPO: 1 minute | RTO: 5 minutes; RPO: 1 minute |
| Site readiness and traffic handling | Geographically separate active sites; automatic traffic redirection | Geographically separate active sites; automatic traffic redirection | Geographically separate active sites; automatic traffic redirection |
| Transaction replication | Synchronous | Asynchronous; maximum lag of 2 minutes | Synchronous |
| Tested full-service recovery | 30 seconds | 30 seconds | 12 hours |
| Maximum committed-transaction loss | Zero | 2 minutes | Zero |
| Meets five-minute RTO? | Yes: 30 seconds | Yes: 30 seconds | No: 12 hours |
| Meets one-minute RPO? | Yes: zero loss | No: 2 minutes of potential loss | Yes: zero loss |
| Decision | Accept: meets both limits | Reject: fails RPO only | Reject: fails RTO only |
Key points
- Remember: fast recovery cannot offset excessive data loss, and zero data loss cannot offset slow recovery.
- Synchronous replication protects committed transactions by persisting them at both sites before acknowledging completion.
- Measure recovery through full-service restoration, not just automatic traffic redirection.
How the exam traps you
- Choose the asynchronous variation because its 30-second recovery meets RTO. Its two-minute data-loss window exceeds the one-minute RPO.
- Choose the slow-recovery synchronous variation because it loses no committed transactions. Its 12-hour recovery exceeds the five-minute RTO.
Question 2
Hint
Focus on the required recovery time after the outage, not just whether data is preserved.
Answer B. A 30-minute recovery requirement implies a very short recovery time objective, so the organization needs processing capability that is already available or can fail over almost immediately. A hot site or redundant failover environment is designed to support rapid restoration of operations with minimal interruption. The other choices require more setup, restoration, or rebuilding time and are unlikely to meet the 30-minute target.
Why not A: A warm site offers some preparedness, but restoring data and completing startup steps typically takes longer than 30 minutes for a mission-critical system. Thus a warm site usually cannot meet a very short RTO like 30 minutes.
Why not C: A cold site provides space and utilities but requires substantial installation and configuration before processing can resume. That setup time makes it unsuitable when availability must be restored within 30 minutes.
Why not D: Backups are important for data recovery but do not provide immediate system availability. Rebuilding the production environment from backups would typically take far longer than the required 30-minute restoration window.
Question 3
Hint
Distinguish the business-determined recovery targets (outcomes) from technical measurements or budgetary limits when deciding which factor should govern the choice.
Answer C. RTO and RPO are business-defined targets that state how quickly systems must be restored and how much data loss is acceptable, so they directly determine which availability architecture is required. Near-zero RTO/RPO typically requires synchronous active-active replication; short RTO with tolerable data loss can be addressed with asynchronous replication and a warm standby; long RTO/RPO can be satisfied by periodic backups and manual restore. Technical metrics (component reliability, performance) and cost inform how to implement and tune the chosen architecture but do not replace the business recovery objectives as the governing requirement.
Why not A: This distractor tempts because performance and capacity needs affect architecture choices and user experience. It fails because throughput and latency targets do not define acceptable downtime or data loss; they are orthogonal design requirements and cannot substitute for RTO/RPO when selecting a recovery architecture.
Why not B: This distractor tempts because hardware reliability and fault-tolerance describe achievable availability and seem directly relevant to choosing architectures. It fails because these are technical measurements used to design and estimate uptime; they inform how to meet the business's RTO/RPO but do not themselves define the business recovery targets that should govern the selection.
Why not D: This distractor tempts because cost constraints determine which options are feasible and often drive practical trade-offs. It fails because budget is a constraint on implementation, not a substitute for the business-defined recovery objectives; the organization should first identify the RTO/RPO it requires, then evaluate which cost-effective options meet those targets.
Question 4
Hint
Calculate the maximum allowable monthly downtime (99.95% of a 30-day month = 21.6 minutes) and compare each architecture's expected unplanned downtime and replication lag against the RTO (30 min) and RPO (5 min) targets.
Answer B. An active-active design with synchronous replication and automatic load balancing produces negligible unplanned downtime per incident (<1 minute) and yields near‑zero RPO, so it satisfies the 21.6‑minute/month availability cap, the 30‑minute RTO, and the 5‑minute RPO. The other architectures either allow replication lag greater than 5 minutes or have per‑incident downtime that would breach the monthly availability threshold.
Why not A: Tempting because automated failover (~8 minutes) meets the RTO and the monthly downtime cap, but asynchronous replication can lag 7-10 minutes, which exceeds the 5‑minute RPO requirement, so it fails the stated RPO target.
Why not C: Tempting because the ~25‑minute manual recovery is within the 30‑minute RTO, but a single 25‑minute outage exceeds the monthly availability allowance (25 > 21.6 minutes given the one‑outage/month assumption) and asynchronous replication lag (up to 10 minutes) exceeds the 5‑minute RPO.
Why not D: Tempting on cost grounds, but restore time (~8 hours) and data loss (RPO = 24 hours) far exceed the 30‑minute RTO and 5‑minute RPO, and the lengthy downtime would violate the monthly availability target.
Question 5
Hint
Separate controls that improve recovery after failure from controls that keep processing available through the failure.
Answer A. The requirements demand near-continuous availability for a single server or OS failure with zero loss of committed transactions. An active-passive cluster with automatic failover and synchronous local replication meets the very short interruption tolerance and ensures no committed transactions are lost on that type of failure. The 8-hour site-outage objective applies to full-site disasters and does not justify leaving a single production node as a local single point of failure.
Why not B: Improving log-backup frequency reduces potential RPO but does not provide automatic continuation of service during a server or OS failure and will not meet the 2-minute interruption objective; it addresses recoverability rather than immediate availability.
Why not C: A warm site with manual cutover is appropriate for site-level disasters where an 8-hour recovery is acceptable, but it does not prevent downtime from a single local server or OS failure and therefore fails the 2-minute and zero-data-loss requirements for that scenario.
Why not D: More testing and longer retention improve confidence in recovery processes but do not remove the single point of failure in production; restore-based approaches are recovery measures, not high-availability solutions that keep processing running through a server or OS failure.
Question 6
Hint
Decide which choice gives both a ready offsite system for rapid failover (short RTO) and a replication SLA that limits data loss to under five minutes (short RPO).
Answer C. A geographically separate hot standby with synchronous or sub‑5‑minute asynchronous replication and automated failover meets both targets: automated failover and a prewarmed standby support an RTO ≤ 1 hour, while the replication SLA limits data loss to within the ≤ 5‑minute RPO. The offsite location protects against site‑level disasters (e.g., flooding) and retaining the cold site preserves cost savings for lower‑priority workloads.
Why not A: This is tempting because it is lower cost and improves over daily tapes, but hourly backups cannot guarantee a ≤5‑minute RPO, and recovery to a cold site (48‑hour capability) cannot meet the ≤1‑hour RTO.
Why not B: A warm site is attractive as an intermediate-cost improvement, yet a 4-6 hour bring‑online time exceeds the ≤1‑hour RTO. Asynchronous replication can also lag and often cannot assure a ≤5‑minute RPO unless specifically configured to do so.
Why not D: Improving local hardware and power redundancy reduces the risk of localized failures and is cost‑effective, but it does not protect against a full site‑level disaster (such as flooding) and therefore cannot reliably achieve the required offsite RTO and RPO targets.
Question 7
Hint
Evaluate both thresholds separately first, then ask which option is the only one that fixes both without depending on rebuild time.
Answer C. The commitment includes both an RPO (≤1 minute data loss) and an RTO (restore within 15 minutes). The current design violates the RPO (5-minute async replication) and the RTO (building plus manual cutover of ~20-24 minutes exceeds 15 minutes). A hot standby that is already running in Region B with sufficiently current replication and automated failover addresses both thresholds simultaneously.
Why not A: This fixes the RPO by reducing potential data loss, which makes it tempting, but it leaves the recovery process manual. Building/activating and performing manual cutover plus DNS changes still push total recovery time beyond the 15-minute RTO, so it is not sufficient.
Why not B: More frequent testing improves preparedness and confidence but does not change the architecture or the actual replication and cutover timings. Testing cannot by itself shorten replication intervals or eliminate manual rebuild and cutover time, so it won't meet the stated thresholds.
Why not D: Local redundancy helps against component or zone failures but does not address a region-wide outage. Backups and rebuild procedures also cannot achieve the ≤1 minute data-loss objective or the 15-minute recovery goal for a regional failure, so this choice misses both requirements.
Question 8
Hint
Compare backup frequency and estimated failover time to the firm's RTO/RPO; the primary issue is the factor that most directly prevents meeting those timing targets.
Answer D. Nightly backups completed at 02:00 allow up to ~24 hours of potential data loss, and the manual failover process requires about 24 hours, together these cannot meet a 1‑hour RPO or a 4‑hour RTO. Therefore the primary availability problem is the warm‑site recovery design (insufficient replication frequency and lack of automated failover). Remediations include near‑real‑time replication and scripted/automated failover or a hotter standby configuration.
Why not A: This option tempts because cross‑site storage replication would support near‑real‑time data consistency and speed failover. It is not the most direct primary issue in the stem, though, since the warm‑site is receiving only nightly backups and requires manual failover, an operational recovery design that alone prevents meeting the 4‑hour RTO and 1‑hour RPO even before detailed storage architecture is addressed.
Why not B: This distractor tempts because network connectivity is a single point of failure that can block replication and failover. It fails as the primary issue here because even with redundant network links the current warm‑site design (nightly backups and manual activation) would still prevent meeting the stated RTO/RPO; network redundancy is important but secondary to the recovery approach.
Why not C: This choice is tempting because an unreliable generator can cause site outages and should be remediated to reduce downtime risk. It is not the primary availability issue relative to the stated recovery objectives, however, because fixing power reliability does not change that recovery to the warm site via nightly backups and manual processes would still take ~24 hours and allow up to a day of data loss.
Question 9
Hint
Match both parts of the SLA: whether the design prevents any data loss (RPO) and whether failover is fast enough (RTO); consider synchronous vs asynchronous replication and whether failover is automated and spans sites.
Answer A. Active-active clustering with synchronous replication and automated failover best satisfies both zero data loss (RPO = 0) and the very short RTO requirement because writes are committed to both sites and client traffic can be redirected immediately. The 20 ms intersite latency makes synchronous replication feasible. This design protects against an entire data-center failure while minimizing interruption.
Why not B: Tempting because synchronous replication satisfies RPO = 0, but manual failover will almost certainly exceed the 30-second RTO requirement; the SLA needs automated failover or immediate client redirection.
Why not C: Tempting because automatic failover addresses RTO, but asynchronous replication can lose in-flight transactions at the time of a site failure, so it does not guarantee RPO = 0 and therefore fails the zero-data-loss requirement.
Why not D: Tempting because on-site redundancy handles hardware faults, but it fails to protect against an entire data-center outage and does not provide the required immediate cross-site failover or zero data-loss.
Common questions
What is the difference between RTO and RPO in disaster recovery?
RTO limits how long service can remain unavailable. RPO limits how much recent data can be lost. QuickShip requires recovery within five minutes and data loss of no more than one minute.
Does asynchronous replication meet a one-minute RPO?
The variation shown does not. Its maximum replication lag is two minutes, so potential data loss exceeds QuickShip's one-minute limit even though recovery takes only 30 seconds.
Does synchronous replication guarantee that a design meets RTO?
No. Synchronous replication addresses committed-transaction loss, not full-service recovery time. The slow-recovery variation loses no committed transactions but fails the five-minute RTO.
Watch it solved
A full CPA ISC task-based simulation on System availability, worked step by step.
System Availability & RTO/RPO | CPA ISC Simulation (TBS) on YouTube
Practice ISC like the real exam
The free ChatCPA simulator: real exam layout, timed testlets, starting with a question on this topic. No account needed to start.
Open the free simulator →More on System availability
- When to Trigger Failover: RTO and RPO
- System Availability: Redundancy and Automatic Failover
- How to Calculate SLA Uptime at 99.95% Availability
- RTO vs RPO: Synchronous vs Asynchronous Replication
- System Availability: Why Backups Do Not Prove Recovery
- Availability vs. Security, Confidentiality, and Integrity
- Availability Calculation: MTBF, MTTR, and Redundancy
Questions from the ChatCPA bank of 17,658 CPA exam questions, each with a written reason for every wrong answer. ChatCPA is built by Nicholas Miller, CPA (Oregon #14907). How these pages are made. Spot an error? Tell us.