AWS Solutions Architect Associate – Domain 3: Design High-Performing Architectures

Vocabulary flashcards covering the key performance, compute, storage, database, and networking concepts required for AWS Solutions Architect Associate Domain 3.

Elasticity

Ability to acquire resources when demand rises and release them when it falls (e.g., EC2 Auto Scaling).

Scalability

Capability of a system to handle increased workload, either by scaling up (vertical) or out (horizontal).

IOPS (Input/Output Operations Per Second)

Metric that measures how many read/write operations a storage device can perform each second.

Latency

Delay between a request and the start of data transfer; lower latency = faster response.

Throughput

Actual rate of data transfer, usually measured in Mbps or Gbps.

EC2 Instance Family Prefixes

G = Graphics, C = Compute-Optimized, R = Memory-Optimized, I = I/O-Optimized, T = Burstable General Purpose.

Burstable (T) Instances

EC2 instances that earn CPU credits while idle and spend them under load; ideal for spiky, low-traffic workloads.

Auto Scaling

Service that automatically adds or removes EC2 instances based on schedule or CloudWatch metrics.

Placement Group – Cluster

Packs instances within one AZ for lowest network latency and highest throughput; used for HPC.

Placement Group – Spread

Places each instance on separate racks to reduce simultaneous failure risk; ideal for critical workloads like databases.

Placement Group – Partition

Distributes instances across partitions (rack sets) within an AZ; designed for large, distributed systems (e.g., HDFS, Cassandra).

Amazon EBS

Network-attached block storage for EC2 that persists independently of instance life cycle.

gp3 Volume

General-purpose SSD EBS volume offering baseline 3,000 IOPS with independent throughput and size settings.

io2 Block Express

Premium SSD EBS volume supporting >16,000 IOPS for mission-critical databases.

st1 Volume

Throughput-optimized HDD EBS volume for frequently accessed, big, sequential data (e.g., log processing).

sc1 Volume

Cold HDD EBS volume for infrequently accessed data at lowest cost.

Amazon EFS

Managed, shared NFS file system that automatically scales to petabytes and can be mounted by thousands of EC2 instances.

Amazon FSx for Windows File Server

Fully managed SMB file system for Windows workloads with NTFS and Active Directory integration.

Amazon FSx for Lustre

High-performance parallel file system for Linux HPC, ML, and media workloads; integrates with S3.

AWS Global Accelerator

Service that uses the AWS global network and Anycast IPs to improve application latency and availability.

Amazon CloudFront

AWS CDN that caches content at edge locations to deliver static and dynamic assets with low latency.

Application Load Balancer (ALB)

Layer 7 load balancer for HTTP/HTTPS with content-based routing.

Network Load Balancer (NLB)

Layer 4 load balancer built for ultra-high performance and static IP addresses.

Object Storage

Flat storage of data as objects with metadata and unique IDs; highly scalable (e.g., Amazon S3).

Block Storage

Low-latency storage presented as raw blocks to a single host (e.g., Amazon EBS).

File Storage

Hierarchical, shared file system accessed with standard protocols like NFS or SMB (e.g., EFS, FSx).

Multi-part Upload

S3 feature that uploads large files in parts for reliability and parallelism; recommended for objects >100 MB.

S3 Transfer Acceleration

Feature that speeds up uploads and downloads to S3 using CloudFront edge network.

S3 Storage Class

Tier (e.g., Standard, Intelligent-Tiering, Glacier) that balances cost vs access frequency.

EBS Snapshot

Point-in-time backup of an EBS volume stored durably in S3.

Instance Store

Ephemeral, physically attached storage that is lost when an EC2 instance stops or terminates.

EFS Performance Modes

General Purpose (low latency) and Max I/O (higher throughput, higher latency).

SMB Protocol

Windows file-sharing protocol used by FSx for Windows.

NFS Protocol

Network File System protocol used by EFS for Linux servers.

Active Directory Integration

Capability of FSx for Windows to join Microsoft AD for authentication and permissions.

Amazon CloudWatch

Monitoring service that collects metrics, logs, and alarms for AWS resources.

Custom Metric – Memory Utilization

RAM usage metric sent to CloudWatch via the Unified Agent; not provided by default for EC2.

Lambda@Edge

Extension that lets you run Lambda functions at CloudFront edge locations for lower latency.

Lambda Max Execution Time

Single Lambda invocation can run up to 15 minutes.

Read Replica

Asynchronous, read-only copy of an RDS/Aurora database used to offload read queries.

Multi-AZ Deployment

Synchronous standby copy of an RDS database in another AZ for high availability and failover.

Connection Pooling (RDS Proxy)

Service that reuses database connections to reduce overhead and improve scalability.

Amazon Aurora

Cloud-native MySQL/PostgreSQL-compatible database with shared storage replicated six times across three AZs.

Aurora Global Database

Aurora configuration that replicates data across regions for disaster recovery and low-latency global reads.

Aurora Serverless

On-demand Aurora cluster that automatically scales ACUs and can pause to zero when idle.

Amazon DynamoDB

Fully managed key-value and document database delivering single-digit millisecond latency at any scale.

DynamoDB Accelerator (DAX)

In-memory cache that reduces DynamoDB read latency to microseconds.

ElastiCache Redis

Managed Redis cluster for in-memory caching with persistence, replication, and failover.

ElastiCache Memcached

Managed Memcached cluster offering simple, multi-threaded in-memory caching.

Amazon VPC

Logically isolated virtual network in AWS where you launch resources; private by default.

Site-to-Site VPN

Encrypted tunnel over the public internet linking on-premises networks to a VPC.

AWS Direct Connect (DX)

Dedicated, private physical connection from your data center to AWS for consistent high bandwidth and low latency.

AWS Transit Gateway

Regional hub that simplifies connectivity between multiple VPCs and on-premises networks (hub-and-spoke).

VPC Gateway Endpoint

Free route-table entry that privately connects a VPC to S3 or DynamoDB.

VPC Interface Endpoint

Elastic Network Interface powered by PrivateLink for private access to most AWS services.

Route 53 Latency Policy

DNS routing that sends users to the region with the lowest network latency.

Route 53 Failover Policy

DNS routing that directs traffic to a primary endpoint and switches to standby on health-check failure.

Route 53 Geolocation Policy

DNS policy that routes users based on their geographic location (country/continent).

Target Tracking Scaling

Auto Scaling policy that adjusts capacity to keep a metric (e.g., CPU) at a target value.

ALB OSI Layer

Application Load Balancer operates at Layer 7 (HTTP/HTTPS).

NLB OSI Layer

Network Load Balancer operates at Layer 4 (TCP/UDP).

S3 Max Single Object Size

5 TB hard limit for any one file stored in Amazon S3.

EBS Max Volume Size (gp3/io2)

Most EBS volume types support up to 16 TiB per volume (64 TiB for io2 Block Express).

EFS Single File Size Limit

A single file in EFS can be up to 47.9 TiB.

FSx for Windows Max Size

Individual FSx for Windows file systems can be provisioned up to 64 TiB.

Kinesis Data Streams

Core real-time streaming service where you manage shards and build custom consumers.

Kinesis Data Firehose

Fully managed service that delivers streaming data to S3, Redshift, or OpenSearch with optional transformation.

Kinesis Data Analytics

Service to run SQL or Apache Flink against streaming data for real-time analytics.

Kinesis Video Streams

Managed service for ingesting and storing real-time video streams.

AWS Glue

Serverless ETL service with a central Data Catalog for discovering, transforming, and loading data.

Amazon EMR

Managed cluster platform for big-data frameworks like Spark and Hadoop on EC2.

Data Lake

Central repository on S3 that stores structured and unstructured data at any scale.

ETL (Extract-Transform-Load)

Process of pulling data from sources, transforming it, and loading it into a destination such as a data lake.

Amazon Athena

Serverless service that runs SQL queries directly against data stored in S3.

Amazon QuickSight

Business-intelligence service for interactive dashboards and data visualizations.

AWS Snowball

Rugged physical device for offline transfer of terabytes-to-petabytes of data to AWS.

AWS DataSync

Online service that moves data between on-premises NFS/SMB and AWS storage at high speed.

AWS Transfer Family

Managed SFTP, FTPS, and FTP endpoints that store files in S3 or EFS.

Database Migration Service (DMS)

Managed tool for migrating databases to AWS with minimal downtime.

ECS EC2 Launch Type

Container tasks run on a customer-managed cluster of EC2 instances.

ECS Fargate Launch Type

Serverless mode where AWS provisions and manages the compute for container tasks.

SQS Queue Depth Metric

ApproximateNumberOfMessagesVisible metric used to scale worker fleets based on backlog size.

CPUUtilization Metric

Default CloudWatch metric showing the percentage of CPU used by an EC2 instance.

GraphQL

A modern query language for APIs that enables clients to request only the data they need, improving efficiency and performance.

AWS AppSync

A fully managed service that allows developers to create GraphQL APIs by handling the real-time data synchronization and connecting to various data sources.

Lazy Loading (Cache-Aside)

A design pattern in which data is loaded into cache only when it is requested, improving performance by loading data on demand rather than preloading it. This pattern helps reduce memory usage and improves speed, as only relevant data is fetched.

Write-Through Caching

A caching strategy where data is written to the cache and the underlying data store simultaneously, ensuring that the cache is always up to date and providing faster read access.