NAVER Cloud Platform Certified Professional flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

Zone

An isolated data center footprint within a region. Placing each application tier across at least two zones lets a single zone failure reduce capacity without taking the service offline.

VPC

A logically isolated private network with its own IP CIDR range. Subnets, route tables, and gateways are defined inside it, and Classic-era resources sit outside it.

Tag-based cost allocation

Attaching key/value tags to every resource so billing reports and monitoring filters can be scoped by environment, team, or project without maintaining manual resource lists.

Login key

The SSH key pair created when launching a server. The public half is stored by the platform and the private half can be downloaded only once, so it must be saved immediately.

Spec type

A fixed bundle of vCPU and memory (standard, high-CPU, high-memory, GPU) chosen at server launch. Changing it requires stopping and starting the server.

NKS node pool

A group of worker nodes in a Kubernetes Service cluster that shares a spec and can be scaled or upgraded independently, ideally spread across multiple zones.

Container Registry

A namespaced repository service for Docker/OCI images. Kubernetes clusters pull images from it using an authentication secret or an attached service account.

Auto Scaling policy

A rule that adjusts group size within min/max bounds using target tracking on a metric, a schedule, or manual action, driven by a launch configuration.

Block Storage

A volume attached to one server in one zone, used for filesystems and database data directories. It is not shared and is not reachable over the network by other servers.

Object Storage

A flat key namespace for blobs accessed over HTTP, governed by bucket policies and storage classes. It is not a filesystem and must not back a live database.

NAS storage

A shared network filesystem mounted over NFS or SMB by several servers at once. Suitable for shared content, but not for high-IOPS transactional database files.

Managed DB failover

A primary database with a standby in another zone; the platform promotes the standby on failure. Applications must use the correct primary endpoint, not the read-only one.

Access Control Group

A stateful virtual firewall attached to servers with allow-only inbound and outbound rules. Traffic matching no rule is denied by default.

Application Load Balancer

A layer 7 load balancer for HTTP/HTTPS that supports host and path routing and TLS termination, forwarding to a target group with health checks.

Network Load Balancer

A layer 4 load balancer for TCP/UDP traffic that preserves the client source IP address and forwards to a target group with health checks.

NAT Gateway

A regional gateway placed in a public subnet that gives private-subnet servers outbound internet access while blocking unsolicited inbound connections.

CDN cache purge

Invalidating cached objects at edge locations after the origin content changes. Without a purge, users keep receiving stale content until the cache-control lifetime expires.

Site-to-Site VPN

An IPsec tunnel between an on-premises gateway and the VPC, using pre-shared keys plus BGP or static routes. Throughput is bounded by the gateway and internet path.

IAM policy union

A principal's effective permission is the union of all policies attached to its user, its groups, and any assumed roles. Adding a group never reduces existing permissions.

Sub-account

A separate identity created under the main account, with its own credentials and least-privilege policies. The main account retains billing and top-level administration.

Key Management Service

A service that stores and rotates encryption keys referenced by Object Storage, Block Storage, and managed databases, so applications never embed raw key material.

Certificate Manager

A service that issues and manages TLS certificates used by load balancers and CDN. Expiry tracking matters because a lapsed certificate causes an immediate outage.

Alarm condition

A monitoring rule combining a metric, a threshold, a sustained duration, and an action such as notification or scaling. Without a duration, brief spikes cause alert noise.

Monitoring agent

Software installed on a server that reports metrics the hypervisor cannot see, such as memory usage, disk utilization, and process counts, to the monitoring service.

Cloud Functions

A serverless runtime executing code on HTTP, schedule, or storage events, billed per invocation. It suits short glue tasks, not long-running stateful jobs.

Idempotent automation

Create-or-update logic in API or CLI scripts so a retried call produces the same end state instead of duplicate resources when a request times out.

AI service API key

A credential authenticating calls to speech, vision, language, and generative AI endpoints. Quotas and rate limits apply per key, so scope keys per application.

At-least-once delivery

A messaging guarantee where a queue or topic may redeliver a message after a failure. Consumers must therefore be idempotent to avoid processing the same event twice.

Data lake layering

Keeping raw immutable data in Object Storage, transforming it into a curated layer, and registering schemas so downstream consumers are not broken by format changes.

Database cutover

Replicating a source database to a managed target, verifying consistency, then switching the application endpoint during a low-traffic window with a tested rollback.

Egress cost

Charges for data leaving the platform to the internet or another region. It is the most commonly underestimated cost, reduced by CDN caching and VPC-internal traffic.

Budget alarm

A billing threshold notification that surfaces spending anomalies early, such as a runaway autoscale group or orphaned resources, before the monthly invoice arrives.