Cloud Architect interview questions

Cloud computing provides on-demand technology resources over a network with metered usage.

• Resources include compute, storage, databases, networking, and higher-level managed services.
• Customers provision capacity through APIs or consoles instead of purchasing and installing all hardware themselves.
• Providers operate large shared platforms, while customers configure and use the resources assigned to them.

These models describe where cloud resources run and who controls the underlying platform.

• Public cloud uses a provider's shared infrastructure, such as AWS or GCP, with customer workloads logically isolated.
• Private cloud dedicates the cloud platform to one organization, often in its own data center or hosted environment.
• Hybrid cloud connects private or on-premises systems with public cloud services under a coordinated architecture.

An IaaS customer manages the operating system and everything above it while configuring provider-supplied compute, storage, and networking.

• The provider manages physical facilities, hardware, and the virtualization layer.
• The customer manages operating systems, patches, runtime software, applications, and most resource configuration.
• Virtual machines such as Amazon EC2 are common IaaS resources.

PaaS removes most server, operating-system, runtime, patching, and platform-scaling work from the application team.

• The provider manages servers, operating systems, runtime components, and much of scaling and patching.
• The customer manages application code, configuration, identities, and data.
• Managed application runtimes and database platforms are common PaaS examples.

A SaaS provider operates the complete application stack, while customers access and configure the finished product over a network.

• The provider operates the application, runtime, infrastructure, updates, and service availability.
• The customer manages users, access settings, configuration, and the data entered into the product.
• Email, collaboration, and customer relationship management products are common SaaS categories.

The shared responsibility model divides security and operational duties between the cloud provider and the customer.

• The provider secures physical facilities, hardware, and the managed layers of each service.
• The customer remains responsible for identities, data, resource configuration, and application security.
• The exact boundary changes by service model, with customers managing more in IaaS than in SaaS.

A cloud region is a geographic area where a provider operates a group of isolated infrastructure locations.

• Customers select a region to place workloads near users or meet latency, residency, and service-availability requirements.
• Regions are connected through provider networks but do not automatically share customer resources or data.
• Service catalogs, quotas, and prices can differ between regions.

An availability zone is an isolated infrastructure location within a cloud region.

• A zone contains one or more data centers with separate power, cooling, and physical failure boundaries.
• Zones in a region have low-latency connectivity for applications that span them.
• Placing replicas in different zones reduces dependence on one facility, but the application must support that layout.

An edge location is a provider site near users where a CDN can serve cached content and related network services.

• The CDN copies cacheable objects from an origin to edge locations after requests or preloading.
• Nearby delivery reduces latency and repeated traffic to the origin.
• Cache keys, expiration, and invalidation control which response can be reused and for how long.

Multi-zone architecture spans failure domains inside one region, while multi-region architecture spans separate geographic regions.

• Multi-zone systems usually have lower inter-site latency and simpler data replication.
• Multi-region systems protect against regional disruption and can place services closer to distant users.
• Multi-region operation adds more complexity in routing, data consistency, deployment, and cost.

A cloud virtual machine is software-defined compute with allocated CPU, memory, storage, and network interfaces.

• The cloud provider runs it on physical hosts through a virtualization layer.
• The customer chooses an operating system image, instance size, network placement, and access rules.
• The customer normally manages operating-system patches and software installed inside the machine.

A machine image is a reusable template used to create virtual machine instances.

• It normally contains an operating system and can include preinstalled packages and configuration.
• Versioned images make instance creation consistent across replacements and environments.
• Images must still be patched, scanned, and rebuilt as their operating system and software age.

Instance families group virtual machines by resource profile, while sizes define the amount of resources in that family.

• General-purpose instances balance CPU and memory for common workloads.
• Compute-, memory-, storage-, and accelerator-optimized families emphasize a particular resource.
• A size specifies values such as virtual CPUs, memory, network performance, and attached-storage limits.

Object storage keeps data as objects addressed by keys inside buckets rather than as files on a server disk.

• Each object contains its data, metadata, and a unique key.
• It scales to large amounts of unstructured data such as images, backups, logs, and artifacts.
• Access uses service APIs, and durability, permissions, versioning, and lifecycle rules are platform features.

Block storage presents a volume of fixed-size blocks that an operating system can format and mount.

• It commonly provides persistent disks for virtual machines and database workloads.
• The operating system manages the filesystem placed on the volume.
• Capacity, throughput, input-output operations, snapshots, and attachment rules depend on the selected volume type.

Managed file storage provides a shared hierarchical filesystem through protocols such as NFS or SMB.

• Multiple clients can access familiar directories and file paths at the same time.
• The provider operates the storage servers, capacity, and availability of the managed service.
• It suits applications that require filesystem semantics and cannot use an object-storage API directly.

Storage classes provide different price, access, and retrieval characteristics, while lifecycle policies move or expire data automatically.

• Frequently accessed classes prioritize immediate access and charge more for ongoing storage.
• Archive classes reduce storage cost but can add retrieval fees and restoration delay.
• Lifecycle rules apply age or access conditions to transition, version, or delete eligible objects.

A virtual private cloud is a logically isolated network space for a customer's cloud resources.

• It defines IP address ranges, subnets, routes, gateways, and network security controls.
• Resources inside it communicate through provider-managed virtual networking.
• Connections to the internet, other virtual networks, or on-premises networks require explicit network paths.

A public subnet has a route for direct internet connectivity through an internet gateway, while a private subnet does not.

• A resource in a public subnet also needs an appropriate public address and security rule to receive internet traffic.
• Private-subnet resources can use controlled outbound access through NAT without accepting unsolicited inbound connections.
• Subnet placement alone does not replace identity, firewall, and application security controls.

A route table tells the cloud network where to send packets for specific destination ranges.

• Each route maps a destination prefix to a target such as a local network, internet gateway, NAT gateway, or peering connection.
• Subnets or network interfaces use route tables according to the provider's association model.
• The most specific matching route normally determines the path for a packet.