ISC · Information Systems and Data Management · 9 practice questions
How to Calculate SLA Uptime at 99.95% Availability
Calculate availability using the contract's measurement period, service boundary, and downtime exclusions. Below: CloudWorks' monthly numbers show why excluding valid maintenance still leaves a breach of its 99.95% uptime commitment.
Try one first
Hint
Exclude properly‑notified scheduled maintenance (check the 48‑hour notice rule), then divide unexcluded downtime by 720 hours to compute the monthly availability percentage and compare to 99.95%.
Answer B. Total monthly hours = 720. The 3.0 hours of scheduled maintenance are excluded because the contract allows up to 4.0 hours of scheduled maintenance to be excluded with at least 48 hours' notice (CloudWorks gave 72 hours). Unexcluded downtime = 1.5 + 0.75 + 1.0 = 3.25 hours, so availability = (720 - 3.25)/720 ≈ 99.5486% (≈ 99.55%), which is below 99.95% and therefore a breach.
Why not A: Tempting because it acknowledges the scheduled-maintenance exclusion, but it misstates the arithmetic/rounding. With 3.25 hours of unexcluded downtime, availability is ≈ 99.55%, not 99.95%, so this choice is incorrect.
Why not C: Tempting for candidates who assume scheduled maintenance always counts as downtime. It's incorrect because the contract explicitly excludes up to 4.0 hours of scheduled maintenance when notice is at least 48 hours, and CloudWorks provided 72 hours' notice, so the 3.0 hours should be excluded.
Why not D: Tempting for candidates who try to reclassify incidents as non-service-affecting. It's incorrect because the stem explicitly states that the failover and backup-restore caused user-visible downtime, so they must be included; including them yields 3.25 hours unexcluded downtime (~99.55% availability), below the SLA.
Numbers to know
| CloudWorks' monthly availability commitmentEvaluate compliance for each calendar month. | 99.95% |
| Monthly measurement denominatorThe 30-day month contains 720 hours. This calculation retains the full monthly denominator. | 43,200 minutes |
| Permitted downtime fraction1 − 0.9995 = 0.0005, or 0.05%. Keep the decimal places straight. | 0.0005 |
| Allowed monthly downtime43,200 × 0.0005. Counted downtime above this amount breaches the commitment. | 21.6 minutes |
| Minimum maintenance noticeAdvance notice is required for the scheduled-maintenance exclusion. | 48 hours |
| Actual maintenance noticeCloudWorks met the notice requirement. | 72 hours |
| Monthly maintenance exclusion capThe contract permits up to 4 hours of qualifying scheduled maintenance. | 240 minutes |
| Excluded scheduled maintenanceThe 3-hour maintenance met the notice requirement and stayed below the cap. | 180 minutes |
| Counted power outageThe 1.5-hour unplanned outage has no stated exclusion. | 90 minutes |
| Counted failover interruptionThe unexpected software-update failover caused user-visible downtime. | 45 minutes |
| Counted backup-restore testThe unannounced test caused user-visible downtime and does not qualify for exclusion. | 60 minutes |
| Total counted downtime90 + 45 + 60 = 195 minutes. Do not add the excluded maintenance. | 195 minutes |
| Actual monthly availability(43,200 − 195) ÷ 43,200 × 100. Approximately 99.55%, below 99.95%. | 99.5486% |
| Downtime above the allowance195 − 21.6 = 173.4 minutes. CloudWorks breached the monthly SLA. | 173.4 minutes |
Key points
- Remember: Compare counted downtime with allowed downtime, not with the length of the maintenance exclusion.
- Convert every outage to minutes before adding or comparing amounts.
- Evaluate this calendar month separately; another month's uptime cannot offset its breach.
- Keep the contract's denominator: this example uses all 43,200 monthly minutes.
How the exam traps you
- Count all 375 outage minutes, including qualifying maintenance. Exclude the 180 maintenance minutes because notice and cap conditions were met.
- Count only the 90-minute power outage because the other interruptions were internal. The failover and backup test caused user-visible downtime, so their 105 minutes also count.
- Treat approximately 99.55% availability as meeting 99.95%. Compare 195 counted minutes with the 21.6-minute allowance. The monthly SLA was breached.
Question 2
Hint
Which choice quantifies how 'bad' downtime would be and how quickly services must be restored, rather than describing load, contracts, or budget?
Answer D. Availability targets should be set based on business impact as identified in a formal BIA; the BIA establishes how critical a system is and quantifies financial, regulatory, or safety consequences and acceptable recovery objectives (RTO/RPO). Technical metrics, vendor SLAs, and budget constraints are implementation and procurement considerations that follow from those business-driven requirements.
Why not A: This distractor tempts because concurrent-user counts drive capacity and perceived responsiveness, but it fails because those metrics inform scalability and performance design, not the acceptable duration of downtime or recovery objectives that define availability targets.
Why not B: Cost is an attractive practical constraint, but it is secondary: availability targets should be driven by business impact, after which a cost-benefit analysis determines which levels of uptime are feasible or justify additional investment.
Why not C: SLAs are tempting because they state contractual uptime commitments, yet they reflect vendor promises rather than the organization's needs; the business should first determine required availability via a BIA and then obtain or negotiate SLAs to meet those needs.
Question 3
Hint
Ask what availability is trying to satisfy: the system being operational and usable when users were told it would be.
Answer D. Availability measures whether a system is operational and usable as promised. Therefore the primary driver for designing availability controls is the commitments in customer contracts, uptime targets, notification requirements for maintenance, and recovery-time commitments, which determine monitoring, redundancy, backup, and recovery priorities.
Why not A: Cost considerations may influence implementation choices, but they should not override contractual availability commitments. The availability objective must be set to meet customer expectations, and then implemented as cost-effectively as possible.
Why not B: This is tempting because sensitive data often requires stronger overall controls, but data sensitivity primarily drives confidentiality and integrity controls. It does not directly set the availability target, which comes from contractual uptime and recovery commitments.
Why not C: Accurate and complete transaction processing relates to processing integrity, not availability. Availability concerns whether the system is up and usable when promised, whereas processing integrity concerns correctness of results.
Question 4
Hint
Focus on the user's ability to access and use the system as committed, not just on whether data stayed intact or internal infrastructure stayed up.
Answer C. Availability is about whether users can access and use the system as committed. Because customers could not log in or submit payroll for four hours and that downtime is within the provider's availability commitment, the provider can be considered to have failed its availability commitment. The lack of data loss indicates processing integrity was preserved, but that does not eliminate the availability impact.
Why not A: This is tempting because transactions were delayed; however, processing integrity concerns accuracy and completeness of processing. Here data were intact and queued transactions processed correctly, so the core issue was that users could not access the service, an availability matter.
Why not B: This distractor relies on a false dichotomy that ties authentication solely to security. When authentication failures prevent user access, availability is affected regardless of whether application servers themselves remained operational.
Why not D: This would be plausible if the third party were explicitly excluded from the SLA, but the question states the identity provider is within the defined system boundary and its downtime is not excluded from the availability commitment. Under those facts, third-party outages still impact the provider's availability obligation.
Question 5
Hint
Convert the weekly and 30‑day downtimes into annual hours and compare which category contributes most to total downtime.
Answer D. Weekly scheduled maintenance is 2 hours × 52 weeks = 104 hours/year, while unplanned outages are about 1 hour × 12 months ≈ 12 hours/year. Since scheduled maintenance constitutes roughly 90% of measured downtime, reducing or eliminating the weekly full‑platform restart will have the largest impact on overall availability. Other measures target much smaller contributors or will not address the dominant planned downtime unless the maintenance approach is changed.
Why not A: Consolidating maintenance is a plausible operational choice, which makes this attractive, but a monthly 8‑hour window totals ≈96 hours/year (8×12), saving only about 8 hours compared with weekly 2‑hour windows (104 hours/year). That modest reduction is far less impactful than eliminating or substantially shortening the weekly downtime as in option A.
Why not B: This is tempting because active‑active redundancy reduces the impact of unplanned single‑site failures, but the stem makes clear scheduled maintenance currently takes all production capacity offline across sites. Without changing the disruptive maintenance process, redundancy will not remove the ~104 hours/year of planned downtime that dominates availability.
Why not C: Improving MTTR is sensible for unplanned outages, which is why this option tempts, but unplanned outages account for only about 12 hours/year versus ~104 hours/year from scheduled maintenance. MTTR improvements would therefore yield a much smaller availability gain than addressing the weekly restart.
Question 6
Hint
Convert the month into total minutes first, then determine how many minutes of downtime 0.1% represents.
Answer C. A 30-day month has 30 × 24 × 60 = 43,200 minutes. At 99.9% availability, allowable downtime is 0.1% of 43,200, which is 43.2 minutes. Actual downtime was 50 minutes, which exceeds 43.2 minutes, so the SLA was not met.
Why not A: This misapplies the 0.1% to a single day instead of the entire 30-day month. The SLA basis is the full month's total minutes, so the correct comparison is 50 minutes versus the allowed 43.2 minutes.
Why not B: This reflects a common shortcut of treating 99% and 99.9% as effectively the same. They are different thresholds; 99.9% is stricter, and here the requirement was not met because 50 minutes of downtime exceeds the allowed monthly downtime for 99.9% availability.
Why not D: This is incorrect because SLAs typically permit a small amount of downtime. The issue is whether the downtime exceeded the permitted amount (43.2 minutes for a 30-day month); limited downtime can still meet a 99.9% target.
Question 7
Hint
Focus first on the exact scope of the availability commitment before reacting to the severity or duration of the incident.
Answer D. Option D describes an unplanned outage within Orion's system boundary that caused actual unavailability. Because the month had no other downtime, the 75 minutes of unavailability exceeds what is allowed under a 99.9% monthly uptime commitment (0.1% of monthly minutes), so this incident is an availability exception attributable to Orion. The fact that failover eventually restored service does not negate that an availability lapse occurred.
Why not A: Tempting because the system was inaccessible and the downtime was longer than the keyed incident, but the commitment explicitly excludes scheduled maintenance that was announced at least 48 hours in advance and performed within the approved maintenance window, so this downtime is not counted as an availability exception.
Why not B: This is tempting because affected customers experienced an inability to authenticate, but the outage was caused solely by a customer-managed identity-provider component outside Orion's system boundary, and the commitment expressly excludes outages caused solely by external customer-managed components.
Why not C: This may feel like an availability problem due to degraded performance, but the stated commitment is for uptime (not a response-time SLA) and users could still access and complete transactions, so it does not clearly constitute an availability exception under the defined commitment.
Question 8
Hint
Focus on the stated commitment and its explicit exception, then ask whether the affected system was actually the one covered.
Answer A. Availability is judged against the contractual commitment and its stated exceptions. The contract allows downtime only for scheduled maintenance announced at least 24 hours in advance; an unexpected outage during normal operations falls outside that exception and therefore indicates the availability commitment was not met.
Why not B: Poor performance can harm users, but the contract commitment addresses availability (being able to use the system). Because customers could still access and complete orders, this scenario represents a performance degradation rather than clear unavailability under the stated contract terms.
Why not C: This is tempting because the system was offline, but the contract explicitly excludes scheduled maintenance that is announced at least 24 hours in advance. Since the downtime was scheduled and announced 2 days earlier (more than 24 hours), it falls within the permitted exception and does not show a breach.
Why not D: The contractual commitment applies to the customer-facing online ordering system. Taking a separate nonproduction test environment offline does not affect the production ordering system's availability and is therefore outside the scope of the stated commitment.
Question 9
Hint
For each calendar month compute allowed downtime = 0.0005 × (days in month × 1,440). Compare that month's allowed minutes to that month's actual downtime, do not sum or average months.
Answer C. 99.95% uptime allows 0.05% downtime each month (0.0005 × monthly minutes). Monthly minutes and allowances: Jan 31×1440=44,640 → allowed 22.32 min (actual 20 → OK); Feb 28×1440=40,320 → allowed 20.16 min (actual 25 → breach); Mar 44,640 → 22.32 min (actual 18 → OK); Apr 43,200 → 21.6 min (actual 10 → OK); May 44,640 → 22.32 min (actual 130 → breach); Jun 43,200 → 21.6 min (actual 35 → breach). Because the SLA is measured per calendar month, any month that exceeds its own allowance is a breach: February, May, and June.
Why not A: Tempting because misreading 99.95% as 99.5% dramatically increases allowable downtime (e.g., Feb allowed ≈40,320×0.005 ≈ 201.6 min), making these outages clearly acceptable. It's wrong because you must apply the SLA value given in the problem (99.95%); assuming a different percentage is unsupported.
Why not B: Tempting because aggregating downtime across a long period and comparing to an annual cap feels like an overall performance check and numerically 238 < ~263 minutes. It's wrong because the SLA is specified per calendar month; months that individually exceed their monthly allowance constitute breaches even if a multi‑month total would fit an annual allowance.
Why not D: Tempting because real‑world monitoring systems sometimes filter brief outages or apply rounding, so February's 25 and June's 35 minutes can seem immaterial. It's wrong here because the stem explicitly states all downtime events count and uptime is measured per calendar month; both February (25 min > 20.16 allowed) and June (35 min > 21.6 allowed) exceed their monthly allowances and are breaches.
Common questions
How much downtime does 99.95% uptime allow in a 30-day month?
Using all 43,200 monthly minutes, the allowance is 43,200 × 0.0005 = 21.6 minutes. CloudWorks' 195 counted minutes exceed that limit.
Does scheduled maintenance count against SLA uptime?
Follow the contract's exclusion conditions. CloudWorks excludes its 180 maintenance minutes because 72 hours' notice exceeds the 48-hour minimum and the maintenance stays within the 240-minute cap.
Do excluded maintenance minutes reduce the SLA denominator?
Use the contract's calculation, not a universal assumption. This CloudWorks example retains the full 43,200-minute denominator and removes qualifying maintenance from counted downtime.
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
- Choose Disaster Recovery Architecture by RTO and RPO
- System Availability: Redundancy and Automatic Failover
- 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.