Cloud computing means accessing computing resources — storage, processing power, software — over the internet from a remote provider's data centre, rather than relying solely on a local device's own hardware. This offers scalability (resources can expand or shrink based on demand, rather than being limited by fixed local hardware) and accessibility (services can be reached from any internet-connected device). Many modern cloud services are distributed systems — made up of multiple interconnected computers working together, often across different physical locations — which improves reliability (if one server fails, others can continue serving requests) but introduces genuine complexity around keeping data consistent and synchronised across all those different locations.
Example
A streaming service handling a sudden surge in viewers for a major live event can automatically scale up by allocating more cloud server resources during the surge, then scale back down afterward — a flexibility that would be far more difficult and expensive to achieve if the company had to own and maintain enough physical servers, sitting mostly idle, to handle only its rare peak demand moments.
Key terms
Cloud computing:
Accessing computing resources over the internet from a remote provider's data centre.
Distributed system:
A system made up of multiple interconnected computers working together, often across different locations.
Questions
1. Cloud computing means accessing computing resources:
Over the internet from a remote data centre
Only from a single local device, with no remote access
Without using any internet connection at all
Only through physical, hand-delivered storage devices
2. Scalability in cloud computing means resources can:
Expand or shrink based on demand
Never change once initially set up
Only ever decrease, with no possible expansion
Have no connection to actual demand
3. A distributed system is made up of:
Multiple interconnected computers working together
A single isolated computer with no connections
No computers at all
Only one physical location with no distribution involved
4. One benefit of a distributed system is:
Improved reliability if one server fails
Guaranteed total failure if any single part has an issue
No possible benefit of any kind
Reduced accessibility for users
5. Cloud services can typically be accessed from:
Any internet-connected device
Only one single, specific physical device
No devices at all
Only devices with no internet connection
6. A streaming service scaling up resources during a surge in viewers is an example of:
Cloud scalability
A complete service failure
Something unrelated to cloud computing
A fixed, unchangeable resource allocation
7. Distributed systems can introduce complexity around:
Keeping data consistent across different locations
Nothing at all; distributed systems are always simpler than single-computer systems
Only physical hardware colour and design
A challenge unrelated to data or synchronisation
8. Why might a business prefer cloud computing over maintaining its own physical servers for handling occasional traffic spikes?
Cloud scalability allows temporarily increasing resources only when needed, avoiding the cost of owning enough hardware to handle rare peak demand at all times
Owning physical servers is always cheaper and more efficient than using cloud computing for handling occasional traffic spikes
Cloud computing provides no genuine cost or efficiency advantage over maintaining dedicated physical servers
Traffic spikes have no real bearing on whether a business would benefit from cloud computing scalability
9. Why might a distributed system be considered more reliable than relying on a single central server?
If one server in a distributed system fails, other servers can continue handling requests, avoiding a complete service outage
A distributed system is always exactly as reliable as depending on a single central server
A single central server is always inherently more reliable than any distributed alternative
The number of servers involved in a system has no bearing on its overall reliability
10. Why might keeping data consistent across multiple servers in different locations be a genuine technical challenge for a distributed system?
Updates made in one location need to be synchronised with others, and network delays or failures can create temporary discrepancies between different copies of the same data
Data across multiple different server locations always stays automatically and perfectly synchronised with zero effort required
Network delays and failures never actually have any effect on how data is synchronised in a distributed system
Maintaining data consistency across different locations is never actually a genuine technical challenge in practice
11. Why might a small start-up choose to rely on cloud infrastructure rather than building and maintaining its own physical data centre?
Cloud computing avoids the very large upfront capital cost and ongoing expertise required to build and maintain physical infrastructure, letting a start-up focus resources elsewhere
Building and maintaining a physical data centre is always cheaper and simpler than using cloud infrastructure for a small start-up
Cloud computing provides no meaningful cost or resource advantage for a small, resource-constrained start-up
The size of a company has no bearing on whether cloud computing or physical infrastructure would better suit its needs
12. Why might automatic scaling in cloud computing help control costs, compared to a fixed amount of always-on infrastructure sized for peak demand?
Resources can shrink back down during quieter periods rather than sitting idle and still being paid for at their peak-demand size around the clock
Automatic scaling always costs exactly the same as maintaining a fixed amount of infrastructure sized for peak demand at all times
Infrastructure sized for peak demand never actually sits idle or incurs any unnecessary cost during quieter periods
The ability to scale resources up and down has no genuine connection to controlling a business's infrastructure costs
13. Why might a company processing sensitive customer data need to carefully evaluate a cloud provider's security certifications and practices before choosing to use their services?
Storing data with a third-party provider means trusting that provider's own security measures, so evaluating their practices helps ensure sensitive data is genuinely protected
All cloud providers offer exactly identical security measures, making any evaluation of their practices completely unnecessary
A company's responsibility for data security completely disappears once data is stored with any cloud provider
Security certifications and practices of a cloud provider have no genuine bearing on how well sensitive data is actually protected
14. Why might accessing cloud services from "any internet-connected device" also introduce security considerations that a purely local system wouldn't need to address in the same way?
Broader, remote accessibility increases the number of potential entry points that need to be properly secured against unauthorised access
Accessing a system from many different devices and locations never actually introduces any additional security consideration
Local-only systems and internet-accessible cloud systems always face exactly identical security considerations
The accessibility of a system from multiple devices has no bearing on its overall security requirements
15. Why might a company using cloud services from a major provider (rather than building everything from scratch) still need to carefully consider where their data is physically stored?
Data location can affect legal jurisdiction, privacy regulations, and latency (how quickly data can be accessed), even though the underlying infrastructure is managed by the cloud provider
The physical location of stored data has no genuine bearing on any legal, regulatory or performance consideration for a company
Cloud providers always store all customer data in exactly the same single physical location with no choice or variation
Legal and regulatory considerations never actually apply to data stored using a cloud computing provider
16. Why might a distributed system need to make deliberate trade-offs between consistency (all copies of data being identical at all times) and availability (the system continuing to respond even during a network issue)?
During a network partition, a system generally cannot guarantee both perfect real-time consistency and continuous availability simultaneously, requiring a considered choice based on the specific application's priorities
Distributed systems can always achieve perfect consistency and full availability simultaneously with no possible trade-off
Consistency and availability are always exactly the same requirement with no meaningful distinction between them
Network issues never actually create any genuine trade-off between a distributed system's consistency and availability
17. Why might relying heavily on a single cloud provider create a strategic risk for a business, sometimes referred to as "vendor lock-in"?
Deep integration with one provider's specific tools and infrastructure can make it costly or technically difficult to switch providers later, reducing a business's negotiating power and flexibility
Using a single cloud provider always provides a business with maximum flexibility and negotiating power, with no genuine risk involved
Vendor lock-in is a purely theoretical concern that never actually creates any real business risk in practice
Switching between different cloud providers is always simple and cost-free regardless of how deeply integrated a business has become with one
18. Why might a global company deliberately deploy the same application across multiple geographically distributed cloud data centres, rather than relying on just one?
Distributing across multiple locations can reduce latency for users in different regions and provide redundancy if one data centre experiences an outage
Deploying an application across multiple geographic locations never provides any genuine benefit over using a single data centre
Latency and redundancy have no real connection to how a global company might choose to structure its cloud infrastructure
A single data centre always provides identical performance and reliability to a multi-region distributed deployment
19. Why might a distributed system need mechanisms for reaching "consensus" (agreement) between different servers about the current state of shared data, especially when some servers might be temporarily unreachable?
Without an agreed process for resolving discrepancies, different servers could end up with conflicting versions of the truth, undermining the system's overall reliability and correctness
Distributed systems never actually need to reach any agreement between different servers about the state of shared data
Temporarily unreachable servers never create any genuine risk of conflicting or inconsistent data within a distributed system
Consensus mechanisms provide no real benefit to how a distributed system maintains reliable, accurate shared data
20. Why might a company running a distributed system need to plan for "graceful degradation" (the system continuing to offer reduced functionality rather than failing completely) when part of the infrastructure goes down?
Designing for partial failure means users can often still access core functionality even during an outage, rather than experiencing a total service collapse from a single point of failure
A distributed system should always be designed to fail completely and totally the moment any single part experiences an issue
Planning for reduced functionality during partial outages provides no genuine benefit to users or the overall reliability of a system
Graceful degradation is a purely theoretical concept that is never actually implemented in real distributed systems
21. Understanding cloud computing and distributed systems mainly helps you to:
Explain how remote, scalable computing resources and interconnected systems support modern digital services
Assume cloud computing has no genuine advantage over maintaining local, physical infrastructure
Ignore the complexity involved in keeping data consistent across a distributed system
Treat distributed systems as always being less reliable than a single central server
Answer key (parent copy)
1. Over the internet from a remote data centre
2. Expand or shrink based on demand
3. Multiple interconnected computers working together
4. Improved reliability if one server fails
5. Any internet-connected device
6. Cloud scalability
7. Keeping data consistent across different locations
8. Cloud scalability allows temporarily increasing resources only when needed, avoiding the cost of owning enough hardware to handle rare peak demand at all times
9. If one server in a distributed system fails, other servers can continue handling requests, avoiding a complete service outage
10. Updates made in one location need to be synchronised with others, and network delays or failures can create temporary discrepancies between different copies of the same data
11. Cloud computing avoids the very large upfront capital cost and ongoing expertise required to build and maintain physical infrastructure, letting a start-up focus resources elsewhere
12. Resources can shrink back down during quieter periods rather than sitting idle and still being paid for at their peak-demand size around the clock
13. Storing data with a third-party provider means trusting that provider's own security measures, so evaluating their practices helps ensure sensitive data is genuinely protected
14. Broader, remote accessibility increases the number of potential entry points that need to be properly secured against unauthorised access
15. Data location can affect legal jurisdiction, privacy regulations, and latency (how quickly data can be accessed), even though the underlying infrastructure is managed by the cloud provider
16. During a network partition, a system generally cannot guarantee both perfect real-time consistency and continuous availability simultaneously, requiring a considered choice based on the specific application's priorities
17. Deep integration with one provider's specific tools and infrastructure can make it costly or technically difficult to switch providers later, reducing a business's negotiating power and flexibility
18. Distributing across multiple locations can reduce latency for users in different regions and provide redundancy if one data centre experiences an outage
19. Without an agreed process for resolving discrepancies, different servers could end up with conflicting versions of the truth, undermining the system's overall reliability and correctness
20. Designing for partial failure means users can often still access core functionality even during an outage, rather than experiencing a total service collapse from a single point of failure
21. Explain how remote, scalable computing resources and interconnected systems support modern digital services