ACA CCP flashcards
ACA CCP · Build recall with a focused review session.
32 cards. Read the term and the definition here. The deck at the top of this page is optional.
- NIST cloud characteristics
- Five essential traits: on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service. These distinguish true cloud from traditional hosting.
- IaaS vs PaaS
- IaaS delivers virtual machines, storage, and networks that the customer configures. PaaS adds a managed runtime and middleware, so the customer deploys code while the provider patches the platform.
- Multi-zone deployment
- Placing resources in separate zones within one region. Zones have independent power and networking, so a zone failure does not take down the whole application.
- Vertical vs horizontal scaling
- Vertical scaling resizes an existing instance (more CPU or memory). Horizontal scaling adds or removes instances. Horizontal scaling better supports elasticity and fault tolerance.
- CAP theorem
- During a network partition a distributed system must sacrifice either consistency or availability. Designers choose based on whether stale reads or failed writes are more acceptable.
- Shared responsibility model
- The provider secures facilities, hardware, hypervisor, and managed-service infrastructure. The customer secures data, identities, access policy, and guest operating system configuration.
- Preemptible ECS instances
- Spot-like ECS capacity offered at a large discount that can be reclaimed by the provider. Suitable for fault-tolerant batch jobs, not for stateful production workloads.
- ECS instance families
- Families are typed by workload: general purpose balances CPU and memory, compute optimized favors CPU, memory optimized favors RAM, and others target storage or GPU needs.
- Security group behavior
- A stateful virtual firewall attached to ECS instances. Rules are allow-only, return traffic is automatically permitted, and one instance may join multiple security groups.
- OSS bucket naming
- Bucket names must be globally unique across all Alibaba Cloud accounts, because the bucket name forms part of the object's public endpoint URL.
- OSS storage classes
- Standard, Infrequent Access, Archive, and Cold Archive. Cost decreases and retrieval time and minimum storage duration increase from Standard toward Cold Archive.
- VPC and vSwitch
- A VPC is an isolated private network defined by a CIDR block. vSwitches are subnets inside it, each bound to exactly one zone, where ECS and RDS are deployed.
- Server Load Balancer
- A managed traffic distributor that spreads requests across backend servers and removes unhealthy ones via health checks. Layer 7 variants route HTTP/HTTPS by URL or host.
- Auto Scaling components
- A scaling group defines which instances to manage, a scaling configuration defines the instance template, and scaling rules or alarms trigger adding or removing capacity.
- RDS read-only instance
- A managed replica of an ApsaraDB RDS primary that serves read queries. It offloads reporting and read-heavy traffic from the primary database.
- RAM role
- An identity assumed by trusted entities such as ECS instances or federated users. It grants temporary credentials so long-term access keys need not be distributed.
- RAM policy structure
- A JSON document with Effect (Allow or Deny), Action, Resource, and optional Condition. An explicit Deny always overrides any Allow statement.
- STS temporary credentials
- Security Token Service issues short-lived credentials bounded by a validity period and a policy. Used for federation and for granting applications scoped temporary access.
- KMS envelope encryption
- A customer master key encrypts a data key, and the data key encrypts the data. Rotating or protecting the master key does not require re-encrypting all data.
- Anti-DDoS and WAF
- Anti-DDoS scrubs volumetric network-layer attack traffic. Web Application Firewall filters HTTP/HTTPS attacks such as SQL injection and cross-site scripting.
- ActionTrail
- A service that records API calls made in an account, including caller identity, time, and source IP, supporting audit, compliance, and forensic investigation.
- Root account protection
- The account owner identity has full privileges. Best practice is to enable MFA, avoid daily use, and perform operations through RAM users and roles instead.
- Network ACL vs security group
- A network ACL filters traffic at the subnet level and is stateless, requiring both inbound and outbound rules. A security group is stateful and applies at the instance level.
- RTO and RPO
- Recovery Time Objective is the maximum tolerable downtime. Recovery Point Objective is the maximum tolerable data loss, expressed as a time span of lost transactions.
- 3-2-1 backup rule
- Keep three copies of data, on two different media types, with one copy stored offsite. Cloud snapshots plus cross-region replication implement this pattern.
- Loose coupling with queues
- Components communicate through a message queue instead of direct calls. Spikes are buffered, consumers scale independently, and a failing consumer does not block producers.
- Blue-green deployment
- Two identical environments run in parallel; traffic switches from blue to green after the new version passes health checks, and rollback is a switch back.
- Canary release
- A new version is exposed to a small percentage of traffic first. If error rates and latency stay acceptable, the percentage is gradually increased.
- Health check role
- Load balancers and Auto Scaling probe endpoints periodically. Failing instances are removed from rotation or replaced, containing faults without manual intervention.
- Right-sizing
- Matching instance CPU, memory, and disk to observed utilization. Over-provisioned instances waste cost; under-provisioned instances cause latency and errors.
- Stateless application tier
- Application servers hold no session state locally; sessions live in a shared store. Any instance can serve any request, enabling horizontal scaling and safe replacement.
- Infrastructure as code
- Defining cloud resources in version-controlled templates rather than manual console actions. It makes environments repeatable, reviewable, and recoverable after failure.