ISC · Information Systems and Data Management · 9 practice questions
System Availability: Redundancy and Automatic Failover
Redundant servers with automatic failover reduce downtime only if the components connecting users to those servers also remain available. Below: follow a retailer's ordering system from a single server to a design that supports failures and rolling maintenance.
Try one first
Hint
Match the control to the specific cause of downtime described in the question, not just to availability in general.
Answer A. This response directly addresses the stated availability problem: planned maintenance on a single application server. Redundant application servers behind a load balancer allow one server to stay online while the other is patched, supporting continuous service. Because the database is already highly available, improving redundancy at the application tier is the best fit.
Why not B: Backups and restore testing improve recoverability after data loss or a catastrophic failure, but they do not prevent planned downtime while a server is taken offline for patching. They address recovery, not continuous availability during maintenance.
Why not C: A hot site addresses disaster recovery for a full-site outage and is more complex and costly than needed here. The problem is brief, recurring downtime during maintenance of a single server, which is better solved by local application-tier redundancy.
Why not D: A UPS protects against power interruptions but does not eliminate planned downtime for OS patching. The described outages result from taking the server offline for maintenance, not from power instability.
Step by step
- Find what stops orders
A retailer's ordering app runs on one server; backups work, but a server failure stops orders. Trace the components required to serve an order, because backups alone provide no alternate live processing capacity.
- Add application capacity
Run at least two application instances in separate availability zones. Ensure the remaining healthy capacity can serve orders when an instance is unavailable.
- Route only to healthy instances
Place the instances behind load balancing with automatic health checks. Stop sending traffic to an unhealthy instance and route requests to healthy instances.
- Protect the load balancer
Use redundant load balancers or a highly available managed load-balancing service. A single unprotected load balancer can make every application instance unreachable.
- Patch in stages
Shift traffic away from one instance before patching and restarting it. Verify its health before returning it to service and moving to the next instance.
- Provide database failover
Use a replicated secondary database with automatic promotion, or a managed high-availability database service. Application redundancy cannot preserve order processing if the only database server fails.
- Remove shared network failures
Provide redundant routers and internet connections with automatic failover. Check that supposedly separate connections do not rely on the same carrier path or conduit.
- Test continued order processing
Regularly test application, database, and network failover, plus planned maintenance switchovers. Confirm that users can place orders and that interruption stays within the retailer's availability requirements.
Key points
- Remember: a second server cannot keep service available if both servers depend on one failed connection.
- Separate availability zones reduce the risk that one zone outage disables every application instance.
- Failover tests must verify that users can still place orders, not merely that a standby server starts.
How the exam traps you
- Choose more frequent backups to keep orders flowing during a server failure. Backups support restoration. Redundant capacity with automatic failover supports continued processing.
- Restart every application instance during the same patch window. Patch in stages so healthy instances remain available to serve users.
- Treat two network connections through one shared conduit as fully independent. Check shared dependencies. One conduit failure can disable both connections.
Question 2
Hint
Focus on the control objective affected by the 5-hour outage, not on whether data could be restored later.
Answer C. Availability concerns whether systems are accessible when needed. Nightly backups preserve data for recovery but do not keep the portal running or provide connectivity during an outage. With only one ISP connection and no tested failover, the 5-hour outage indicates an availability control weakness.
Why not A: This reflects a common shortcut of equating backups with availability. Backups support recovery from data loss but do not ensure continuous access or quick failover during an outage. A 5‑hour interruption shows availability was not adequate.
Why not B: This is tempting because an outage disrupts processing, but integrity refers to accuracy and completeness of data. The facts state no data was lost or altered and integrity controls are otherwise effective, so the primary issue is loss of access (availability), not data integrity.
Why not D: Offsite backups can raise confidentiality concerns, but the question states backups are encrypted and gives no indication of unauthorized access or disclosure. The event described is system downtime, which is an availability issue rather than a confidentiality failure.
Question 3
Hint
Focus on the control that best keeps the system usable or restores it quickly after disruption, rather than the controls that protect data or improve processing quality.
Answer D. System availability concerns whether the platform is accessible and usable when promised. A tested failover site and regular recovery drills directly reduce downtime and improve the organization's ability to restore service after an outage, so this control most directly supports the availability commitment.
Why not A: Encryption is an important control for protecting confidentiality of data in storage and transit. However, encryption does not by itself keep the ordering platform operational or reduce downtime, so it does not most directly support availability.
Why not B: Automated edit checks and exception reporting improve processing integrity by helping ensure transactions are valid and accurate. They improve data quality and control effectiveness within the system but do not by themselves ensure the platform stays available during outages.
Why not C: Periodic access reviews are a good security practice that limit inappropriate or excessive access. They primarily address security and misuse risk rather than the system's ability to remain available or to be restored quickly after a disruption.
Question 4
Hint
Focus on the specific cause of the outage and choose the control that keeps the system operating when that same cause happens again.
Answer C. Adding redundant production servers with automatic failover directly removes the single point of failure in the production environment. If one server hardware fails, traffic can be routed to the redundant server with minimal disruption, which directly improves availability. Because the stem rules out security or software-release causes, this is the most effective single investment to prevent a similar outage.
Why not A: Encrypting data protects confidentiality and helps with compliance, yet it does not prevent or reduce downtime caused by a failed production server. Encryption does not provide redundancy or failover capability to maintain availability.
Why not B: Stronger password complexity improves security by reducing the risk of unauthorized access, but it does not address hardware failure or eliminate the single point of failure that caused the outage. It is unrelated to keeping the system running when the production server fails.
Why not D: Formal change approvals can reduce outages caused by bad deployments and are a plausible distractor, but the question specifies the outage was not due to a software release. Change management does not directly prevent hardware failures or provide failover for a single production server.
Question 5
Hint
Which change both removes a single point of failure and lets you patch without taking the whole service offline?
Answer D. An active‑active multi‑AZ deployment behind a load balancer removes the single point of failure at the host or AZ level: if one instance or AZ fails, traffic shifts to healthy instances in other zones. Health checks stop routing to unhealthy instances, and rolling updates let you patch instances one at a time so planned maintenance does not take the whole service offline. These changes directly reduce both unplanned and planned downtime and are the most effective step toward meeting a tight (99.95%) monthly availability SLA.
Why not A: Better backups and more frequent restore testing improve recoverability and reduce data loss, which is tempting for disaster planning, but restore‑based recovery typically takes longer than failover and does not prevent live service downtime from host failures or patch reboots. Thus backups alone will not substantially reduce the service downtime needed to meet 99.95% availability.
Why not B: This is tempting because adding a second instance and using rolling updates addresses host‑level failures and planned reboots, but keeping both instances in the same availability zone preserves an AZ‑level single point of failure. An AZ outage or zone‑level issue could still take the entire service down, so this approach is unlikely to reliably meet 99.95% availability.
Why not C: Improved monitoring and faster response are valuable and will shorten outage durations, which is why this option is attractive, but they do not eliminate the underlying single point of failure or enable non‑disruptive patching. Even with rapid MTTR, host or AZ failures and required full‑VM reboots will still cause outages that are unlikely to meet the tight SLA on their own.
Question 6
Hint
Focus on the control that keeps the service running when one component fails, not the control that helps recover later.
Answer A. Deploying multiple application servers behind a load balancer with health checks gives immediate redundancy and automatic rerouting when one server fails. This approach maintains service availability with little or no interruption, directly meeting the requirement for customers to continue placing orders. It is an availability/high-availability design rather than a post-failure recovery measure.
Why not B: Backups are tempting because they're important for protecting data, but they support recovery after data loss rather than keeping an application running during a server failure. Increasing backup frequency improves recoverability, not real-time availability or automatic failover.
Why not C: A cold site is a disaster-recovery option that can restore operations after a major outage, but activation typically takes many hours. It is too slow to meet the requirement of minimal interruption when a single server fails and does not provide immediate failover.
Why not D: Daily log review helps detect and investigate issues but is a manual, after-the-fact control. It does not provide redundancy or automatic rerouting to maintain service in real time when an application server fails.
Question 7
Hint
Focus on the difference between keeping a system running during a failure and recovering it after a failure.
Answer D. Availability in this scenario is best achieved through redundancy and automatic failover so that a single server failure does not interrupt service. The stem requires little to no interruption and no manual restore, which points to a high-availability design rather than recovery after the fact. Accounting for network and power components prevents hidden single points of failure.
Why not A: Backups are important for recovering data and restoring systems after an outage, but they do not keep a system running with little to no interruption; nightly backups are a recovery measure, not a high-availability design.
Why not B: A hot site provides an alternate facility that can shorten recovery time for major outages, but it typically addresses broader facility-level or regional failures and may require failover procedures; the stem focuses on a single-server failure and no manual transition, which is better solved by in-place redundancy and automatic failover.
Why not C: Encryption addresses confidentiality (and can support integrity), but it does not prevent or hide hardware failures or provide continuous service during a server outage; it is not an availability control for the scenario described.
Question 8
Hint
Focus on the control that keeps the service running during a failure, not the one that only protects data or helps recover later.
Answer A. System availability is about keeping services accessible when needed. Redundant servers combined with automatic failover reduce downtime from hardware or server failures and enable the portal to continue operating with minimal interruption. This directly addresses management's requirement for continued access during a server failure.
Why not B: Backups are important for recovery and restoring data after a failure, but they do not keep the portal available while the production server is down. Backups help restore service later, not provide immediate continuity.
Why not C: MFA strengthens access control and reduces risks from compromised credentials, but it does not provide redundancy or reduce downtime from a hardware or server failure. It improves security, not availability.
Why not D: Encryption protects data confidentiality at rest, which is important, but it does not ensure the portal remains accessible if the single production server fails. Availability requires redundant infrastructure or failover mechanisms.
Question 9
Hint
Focus on which control objective is affected when a single point of failure is removed from system access.
Answer D. Adding a second ISP and automatic failover addresses availability by removing a single point of network failure. If one connection goes down, traffic can be routed through the other, reducing the likelihood of downtime. Because the scenario specifies no changes to access controls, application logic, or encryption, improved availability is the most direct effect.
Why not A: This distractor confuses confidentiality with availability. Failover keeps systems reachable but does not implement controls (like encryption or access restrictions) that prevent inappropriate disclosure of data.
Why not B: This is tempting because students sometimes assume infrastructure changes equal security improvements. However, redundant connectivity and failover do not change authentication or authorization mechanisms, so they do not directly reduce the risk of unauthorized access.
Why not C: While greater availability can reduce missed transactions from outages, processing accuracy and completeness are matters of application design and processing controls (integrity). Network redundancy does not guarantee correct or complete processing of orders.
Common questions
Do backups prevent downtime when a production server fails?
No. Backups support recovery after failure, but the retailer cannot process orders while its only server is unavailable. Redundant servers with automatic failover provide alternate processing capacity.
Is a load balancer enough to ensure system availability?
No. It needs healthy redundant application instances to receive traffic, and the load balancer itself must not remain a single point of failure. Database and network dependencies also need protection.
How do rolling updates reduce maintenance downtime?
They take only part of the application fleet offline at a time. The retailer routes orders to healthy instances while another instance is patched and checked before returning to service.
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
- 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.