C-CATODSW key terms

On-demand delivery of computing resources such as servers, storage, and databases over the internet, with pay-as-you-go pricing and minimal upfront commitment or manual provisioning effort.

The ability to automatically acquire or release resources as demand rises or falls, so capacity tracks workload needs closely without long-term over-provisioning or manual intervention.

The capacity of a system to handle growing load by adding resources, either vertically by enlarging a single node or horizontally by adding more nodes to a pool.

Adding more instances to a pool so work is distributed across them, often paired with a load balancer and stateless application design to spread traffic evenly.

Increasing the size or power of a single resource, such as moving a database to a larger instance class, which usually requires downtime and has a fixed upper limit.

Designing systems to remain accessible despite component failures, typically by deploying redundant resources across multiple zones and using health checks with automatic failover.

The property of a system that keeps operating correctly when one or more components fail, often through full redundancy so no service interruption is visible to users.

A geographic area containing multiple isolated data center groups, chosen to place workloads near users, meet data residency rules, and support cross-region replication for recovery.

One or more discrete data centers within a region, with independent power and networking, used together to build resilient architectures that survive a single-site failure.

The delay between a request and its response, affected by physical distance, network hops, and processing time; keeping workloads near users usually reduces it.

A division of duties where the provider secures the cloud infrastructure itself, while the customer secures data, identities, configurations, and workloads running in it.

The service and practices for creating users, groups, and roles, then granting least-privilege permissions through policies so only authorized principals can act on resources.

Granting only the permissions needed to perform a task, nothing more, to limit the blast radius of mistakes and the impact of compromised credentials.

An identity with permission policies that trusted principals can temporarily assume, providing short-lived credentials instead of long-term keys embedded in code.

Trusting an external identity provider so users sign in with existing corporate or social credentials and receive temporary cloud access mapped to their group membership.

Requiring two or more independent verification factors, such as a password plus a one-time code, so a stolen password alone cannot grant account access.

Protecting stored data by converting it to ciphertext on disk, using keys managed by the provider or the customer, so stolen media reveals nothing readable.

Protecting data moving across networks with protocols such as TLS, preventing eavesdropping or tampering between clients, load balancers, and backend services.

A managed service for creating, storing, rotating, and auditing cryptographic keys, integrating with storage and database services so encryption is applied without custom code.

A service that stores credentials such as database passwords and API keys, controls access to them, and can rotate them automatically on a schedule.

A stateful virtual firewall attached to instances that allows or denies inbound and outbound traffic by protocol, port, and source, evaluated as allow rules only.

A stateless subnet-level filter with numbered allow and deny rules evaluated in order, providing a coarse secondary layer of network control beyond security groups.

A logically isolated network in the cloud where you define address ranges, subnets, routing, and gateways to control how resources communicate internally and with the internet.

A range of IP addresses within a virtual network, labeled public or private depending on routing, used to separate tiers and control which resources are internet-reachable.

A service that distributes incoming requests across multiple healthy targets, performs health checks, and terminates TLS, improving availability and smoothing traffic spikes.

Automatically adjusting the number of running instances based on metrics or schedules, replacing unhealthy nodes and matching capacity to demand to control cost.

A predefined combination of virtual CPU, memory, storage, and networking capacity for a compute instance, chosen to match workload profile such as general purpose or memory optimized.

A model where the provider manages servers entirely, runs code in response to events, scales automatically, and bills only for execution time and requests consumed.

A serverless compute offering where you deploy individual functions triggered by events or HTTP calls, with no server management and automatic per-request scaling.

Storing data as objects in flat buckets, each with a key, metadata, and unique identifier, ideal for unstructured content, backups, and static assets accessed over HTTP.

Persistent volume storage presented to an instance as a raw disk, formatted with a filesystem, offering low latency for databases and operating system drives.

A tier defining durability, availability, retrieval latency, and price for stored objects, letting you trade access speed against cost for infrequently used data.

Rules that automatically transition objects to cheaper storage tiers or delete them after a set age, reducing cost and meeting data retention requirements.

A network of edge locations that caches content close to users, reducing latency and origin load while lowering data transfer costs for popular assets.

A design where producers emit events to a queue, topic, or stream and consumers react asynchronously, decoupling services and absorbing traffic bursts.

A buffer that holds messages until consumers retrieve them, decoupling components, smoothing load spikes, and enabling retries when downstream processing fails.

Billing based on actual consumption of compute, storage, and transfer, with no upfront commitment, so cost scales with usage and experimentation stays inexpensive.

A commitment to use a specific instance family in a region for one or three years in exchange for a significant discount versus on-demand rates.

A flexible commitment to spend a fixed hourly amount for one or three years, applying discounts across eligible compute services regardless of instance family or region.

Spare capacity offered at a steep discount that can be reclaimed with short notice, suitable for interruption-tolerant batch, rendering, or stateless workloads.

The full lifetime cost of a solution, including compute, storage, transfer, licensing, staffing, and migration, used to compare cloud adoption against on-premises operation.

Charges for moving data between regions or out to the internet, while many inbound transfers are free; architecture choices strongly influence this bill component.

A configured threshold that notifies owners when forecast or actual spending crosses a limit, enabling early correction before monthly bills exceed expectations.

A key-value label attached to resources so spending can be grouped by project, team, or environment in billing reports for accountability and chargeback.

A contract stating measurable commitments such as uptime percentage and the remedies, usually service credits, available when the provider misses those targets.

Plans and replicated resources that restore service after a major outage, characterized by recovery time and recovery point objectives that set acceptable data loss and downtime.

A point-in-time copy of a volume or database stored durably, used to restore data, clone environments, and support backup schedules across regions.