ICT-Platform Dev Specialist FA flashcards

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

CAP theorem

A distributed data store can guarantee at most two of consistency, availability, and partition tolerance. During a network partition, the system must choose between rejecting requests (consistency) or serving possibly stale data (availability).

Saga pattern

A way to manage distributed transactions as a sequence of local transactions, each paired with a compensating action. If a step fails, previously completed steps are undone via their compensations, preserving eventual consistency without distributed locks.

Idempotency key

A client-generated unique token sent with a request so the server can detect duplicates. The server stores the result keyed by the token and returns it on retry, preventing double execution of non-idempotent operations such as payments.

Circuit breaker states

Closed allows normal traffic; Open fails fast for a cooldown period after a failure threshold; Half-Open permits limited probe requests. Successful probes close the circuit, while any failure reopens it to protect the downstream dependency.

Twelve-factor app

A methodology for building portable cloud-native services: configuration in environment variables, stateless processes, disposability, dev/prod parity, logs as event streams, and admin tasks run as one-off processes. It improves deployability and operational consistency.

Constructor injection

A dependency injection style where required collaborators are passed as constructor parameters. It makes dependencies explicit, enforces them at object creation, and simplifies unit testing by allowing test doubles to be supplied directly.

Async/await event loop

Async functions suspend at await without creating OS threads; the event loop resumes them when the awaited operation completes. Blocking calls inside async code stall the loop, so use thread pools or async-native libraries for blocking work.

Deadlock four conditions

Deadlock requires mutual exclusion, hold-and-wait, no preemption, and circular wait simultaneously. Breaking any one condition, such as acquiring locks in a global order to eliminate circular wait, prevents deadlock.

Feature flag fallback

A configuration toggle that controls feature exposure independently of deployment. Evaluation must be fast and failure-safe: if the flag service is unreachable, return a safe default rather than blocking or failing the request.

Third normal form

A relational schema is in 3NF when it is in 2NF and no non-key attribute depends transitively on another non-key attribute. Each non-key column depends only on the primary key, reducing update anomalies.

Isolation levels

Read Uncommitted permits dirty reads; Read Committed prevents them but allows non-repeatable reads; Repeatable Read prevents non-repeatable reads but may allow phantoms; Serializable prevents all anomalies, typically using predicate locking or serializable snapshot isolation.

Covering index

An index that contains every column referenced by a query, so the database can answer from the index alone without visiting the table. It reduces I/O and latency for read-heavy workloads at the cost of index size and write overhead.

Consistent hashing

A partitioning scheme that maps keys and nodes onto a hash ring. Adding or removing a node remaps only its adjacent keys, minimizing data movement. Virtual nodes improve balance when physical nodes are few.

Event sourcing

Persisting state as an append-only sequence of domain events rather than current-state rows. Current state is derived by replaying events; snapshots bound replay time, and versioned event schemas support evolution.

Cursor-based pagination

Pagination that returns an opaque token encoding the last seen sort key. Unlike offset pagination, it remains stable when rows are inserted or deleted and uses indexes efficiently, making it suitable for large or frequently changing datasets.

gRPC field numbers

In Protocol Buffers, field numbers identify fields on the wire and must never be reused or renumbered. Removed fields should be marked reserved. Changing a field number breaks compatibility with existing serialized messages.

Outbox pattern

Write domain changes and an event record in the same database transaction, then a relay publishes the event asynchronously. This guarantees at-least-once publication without distributed transactions, and consumers must be idempotent.

OAuth PKCE

Proof Key for Code Exchange adds a code verifier and challenge to the authorization code flow. It protects public clients such as mobile apps and SPAs from authorization code interception, and is required for those clients.

Terraform remote state

Terraform stores resource state in a backend such as object storage with locking. Remote state enables team collaboration and prevents concurrent applies from corrupting state. Locking is typically implemented with a database table or lock service.

Kubernetes readiness probe

A probe that determines whether a container is ready to receive traffic. Failing readiness removes the pod from service endpoints without restarting it. Liveness probes instead restart unhealthy containers and must not check external dependencies.

Blue/green deployment

Running two identical environments, blue and green. New versions deploy to the idle environment, are verified, then traffic is switched. Rollback is a traffic switch back, minimizing downtime and release risk.

Multi-stage container build

A Dockerfile technique using multiple FROM stages. Build tools and dependencies stay in an early stage, and only the runtime artifacts are copied into the final image, reducing size and attack surface.

OIDC ID token

A signed JWT issued by an OpenID Connect provider that asserts user identity. It contains issuer, subject, audience, expiration, and nonce claims. Relying parties must validate the signature and these claims before trusting it.

RBAC versus ABAC

Role-based access control grants permissions through roles assigned to users. Attribute-based access control evaluates policies over user, resource, and environment attributes. RBAC is simpler; ABAC is more granular and context-aware.

Envelope encryption

Encrypting data with a data key, then encrypting that data key with a key management service master key. The encrypted data key is stored alongside the ciphertext. Rotation of the master key does not require re-encrypting all data.

Least privilege

Granting only the permissions required to perform a task, scoped to specific resources and conditions. Use temporary credentials and narrowly scoped policies. It limits the blast radius of compromised identities.

Error budget

The allowed amount of unreliability derived from an SLO, calculated as 1 minus the SLO target over a window. It quantifies how much failure is acceptable and guides release velocity versus reliability investment decisions.

Four Golden Signals

Latency, traffic, errors, and saturation. These four metrics form a default monitoring baseline for request-driven services and help detect both user-visible failures and resource exhaustion before they cause outages.

Tail latency percentiles

p95 and p99 latency values capture the slowest requests, which averages hide. Tail latency dominates user-perceived performance in fan-out systems, so alerting and SLOs should target high percentiles rather than means.

Retry budget

A cap on retries as a fraction of total requests, such as ten percent. It prevents retry storms from amplifying outages. Combine with exponential backoff and jitter, circuit breakers, and server-side load shedding.

Test pyramid

A testing strategy with many fast unit tests, fewer integration tests, and few end-to-end tests. The shape keeps feedback fast and flakiness low; contract tests can replace some end-to-end coverage.

Semantic versioning

A versioning scheme MAJOR.MINOR.PATCH. MAJOR increments for breaking changes, MINOR for backward-compatible features, PATCH for backward-compatible fixes. Pre-release and build metadata are appended with hyphens and plus signs.

DORA metrics

Four delivery metrics: deployment frequency, lead time for changes, change failure rate, and time to restore service. They balance speed and stability and are used to benchmark and improve software delivery performance.