System Architecture and Production Troubleshooting Review

Practice flashcards covering microservices architecture, AWS services (EKS, Lambda, EMR), Spring Boot optimization, and production reliability strategies.

Lambda Cold Start

A source of latency that occurs during initial function invocation; it is a primary reason to bypass Lambda in favor of EKS for requests requiring extremely low latency, typically under $10\,ms$.

Kafka Buffering

The practice of using Kafka as a decoupling layer to absorb traffic bursts and smooth spikes when a system receives high throughput (e.g., $100\,TPS$) that exceeds the capacity of downstream services (e.g., $25\,TPS$).

Exponential Backoff

A retry mechanism implemented alongside dead-letter queues (DLQ) to handle failed messages gracefully by progressively increasing the wait time between retries.

Eclipse Memory Analyzer Tool (MAT)

A tool used to analyze heap dumps generated during OutOfMemoryError (OOM) events to diagnose issues like database connection leaks or unreleased in-memory caches.

Horizontal Pod Autoscaler (HPA)

A sophisticated scaling strategy in EKS that utilizes custom metrics to enable efficient resource utilization and responsiveness to workload changes.

Spring Ahead-of-Time (AOT) Compilation

An optimization technique used to reduce JVM startup overhead and improve containerized deployment agility, helping reduce startup times from approximately $60\,s$ to $15\,s$.

CompletableFuture API

A Java API used to execute tasks, such as fraud detection rules or downstream service calls, in parallel to maximize concurrency and minimize total API latency.

Idempotency Key

A unique identifier, often a transaction ID or request hash, used in payment systems to prevent duplicate processing and double charging during retries.

Amazon EMR with Spark

A distributed computing platform preferred over Lambda for processing large-scale batch datasets (e.g., terabyte-scale or $10\,million$ transaction CSVs) due to Lambda's $15$-minute execution limit.

Aurora Global Database

A service used in multi-region active-active architectures to handle data replication across different geographical regions like the US and Canada.

Route 53 Latency-Based Routing

A DNS routing policy used to direct users to the nearest AWS region to ensure high availability and low latency in a multi-region system.

JWKS Endpoint Caching

A strategy to reduce API latency by localizing public keys used for JWT signature verification, thereby avoiding repeated external calls to the authentication server.

Signature-First Processing Flow

A security protocol where a digital signature is verified using a public key before decrypting data with a private key to prevent the processing of tampered or malicious data.

Bulkhead Pattern

An architectural pattern used to isolate failures within specific services, preventing cascading delays across the system when multiple downstream services are involved.

Circuit Breaker

A resilience mechanism that stops calls to a failing service after a certain failure threshold is reached, instead providing a fallback response to maintain overall system health.

Terraform Remote State

A feature that stores Terraform configuration state in S3, allowing for team collaboration, version control, and reproducible infrastructure environments.

Least Recently Used (LRU)

A cache eviction policy used to optimize memory usage by removing the items that have remained unaccessed for the longest period.

AWS X-Ray

A distributed tracing tool used to visualize request flows and correlate logs with traces to pinpoint bottlenecks and debug production latency issues.