This cheatsheet provides a list of various services provided by Amazon Web Services.
Services mentioned are based on the notes I made for my AWS Cloud practioner exam.
IaaS offering
- Families
- General Purpose: Balanced resource, diverse workloads, webservers, code repos
- Compute Optimized: Compute intensive tasks, game servers, scientific, high performance computing
- Memory Optimized: Processing large data sets in memory
- Accelerated Computing: Floating point calculations, data pattern matching, graphics, used hardware accelerators
- Storage Optimized: High performance for locally stored data
- Pricing
- On demand: No commitment, no bound, billed per second
- Savings: Save upto 72% by 1 or 3 year commitment
- Reserved Instances: Upto 75% savings, suited for predictable steady usage, 1-3 year
- Spot Instances: Upto 90% savings by using spare EC2 instances, 2 minute warning given before shutting down(reclaim period), for flexible (batch) workloads
- Dedicated Hosts: For maintaining compliance, unshared resources
- Features
- EC2 autoscaling: Free feature to add/remove EC2 instances in response to demands.
- Dynamic scaling responds to demands
- Predictive scaling auto schedules based on predicted demand
- EC2 autoscaling: Free feature to add/remove EC2 instances in response to demands.
- Features
- Highly Available
- Auto scalable
- Other instances just know about 1 ELB handling scale
- Send, receive, store messages at any volume
- Payload protected until delivery
- Send notifications to users
- Uses pub/sub model
- Can create topics, subscribe to topics
- Subscribers can be SQS queue, lambda, webhooks
FaaS
- Upload code to lambda function
- Configure trigger
- Code run on trigger
- Features
- Scalable
- Code run under 15 minutes
- Containers run in isolation in EC2 instance as host
- Serverless for ECS and EKS
EC2 for traditional apps requiring full OS access
Lambda for short running functions, service created apps, event driven apps, apps without server provisioning(serverless)
EKS/ECS for docker containers on managable EC2 instances
Fargate for containers on managed instances
Choosing region:
- Compiance
- Latency
- Feature availability
- Pricing
Quantum computing
- CDN service
- Features
- Deliver data, apps and APIs with low latency and high transfer speed
- Uses edge locations
- DNS service
- Routing Policies
- Latency based
- Geolocation based
- Geoproximity
- Weighted round robin
- Mini AWS region in your data center
- Quickly deploy and manage apps in EC2 without worrying about infrastructure
- Infrastucture as Code (IaC)
- Declarative Cloud Formation templates (JSON and YAML)
- Resources like EC2, DB,analytics and ML
- Same template can be run in multiple regions
- Logically isolated AWS cloud
- Subnets allow to group resources
- IGw: Internet Gateway
- Allows incoming traffic
- Virtual Private Gateway
- VPN connection between AWS and OnPremise
- AWS Direct Connect
- Dedicated private connection between AWS and on-premises
- ACL: Access Control List
- Every packet crossing boundaries checked against ACLs
- Stateless
- Allows both inbound and outbound traffic
- SG: Security Group
- Allow/Deny traffic (like allow HTTPS, deny OS requests)
- Stateful
- Does not allows traffic by default
flowchart LR
subgraph subnet2
direction RL
e2((EC2 instance))
f2((gateway))
e2 -->|SG| f2
end
subgraph subnet1
direction LR
e1((EC2 instance))
f1((gateway))
e1 -->|SG| f1
end
subnet1 --->|ACL| subnet2
subnet2 --->|ACL| subnet1
Storage types:
- Block level storage:
- Files stored in series of blocks
- Only updated pieces of block(s) are updated if file update
- Instance store volumes:
- Storage attached to EC2 instance
- Can be lost in case EC2 host is restarted
- Restarted EC2 hosts can start with different volume attached
- Virtual hard drives attached to EC2 instances (not tied to EC2 instance)
- Upto 16 TiB storage, Solid State by default
- Allows incremental backups(snapshots)
- Data store
- Features
- Unlimited amount of data
- Data stored as objects in buckets, file is a complete object
- Maximum 5 TB size for object
- Objects can be versioned
- Different buckets across regions to limit classes
- Tiers
- S3 standard: For static web hosting, 99.999999999 (Eleven 9s)
- S3 IA: S3 Standard Infrequent Access, for backups or archives
- S3 Glacier: Long storage for audit purpose, direct upload using Lifecycle Policies between tiers
- Managed shared File System
- Accessible by multiple instances at same time
- Autoscalable
- Linux File System
- Regional resource (stored data across multiple Availability Zones)
- Do not require to be in same AZ like EBS
- Unlike EBS, EFS is not attached to EC2 instances
AWS supports MySQL, PostreSQL, Oracle and MSSQL Server databases Databases can be migrated using Lift and Shift.
- Databases
- Aurora
- PostgreSQL
- MySQL
- MariaDB
- Oracle DB
- MSSQL Server
- Supports existing database features and provides auto patch, backups, recovery, failovers, failovers, disatser recovery
- Managed MySQL and PostgreSQL Database
-
$\frac{1}{10}$ th cost of commercial database - Upto 15 read replicas and 6 copies of data
- Continuous S3 backups
- Point in time recovery
- Fully managed, serverless, non-relational, NoSQL database
- Data queried using keys, simple, single tabled queries
-
$ms$ response time
flowchart
table[TABLE]
item1[ITEM]
item2[ITEM]
attribute1[ATTRIBUTE]
attribute2[ATTRIBUTE]
table --> item1
table --> item2
item1 --> attribute1
item1 --> attribute2
Data Warehouses: Specifically engineered for big, historical analytics instead of operational analysis (real time)
- Data Warehouse as a service
- Multiple PetaBytes nodes
- Query against exabytes of data
- Migrate existing databases securely and easily
- Source DB is operational during migration to minimize downtime
- Source and target DB not needed of same type
- Homogenous DB: Same type
flowchart LR
mysql[MySQL]
rds[Amazon RDS]
oracle[Oracle]
oraclerds[Oracle RDS]
mysql <--> rds
oracle <--> oraclerds
- Source can be on-prem, EC2 or RDS
- Target can be EC2 Database or RDS
- Heterogenous migrations need to convert schema structures and database codes using AWS Schema Conversion Tool before DMS conversion
- Dev and Test migrations: Copy production data to dev or test database
- Database consolidation: Combine databases to common database
- Continuous DB replication
- Dynamo DB: Key/value pair
- Document DB: MongoDB compatibilty
- Neptune: Graph database for social networking, recommendation engies, fraud detection
- QLDB: Quantum Ledger DataBase, immutable DB
- Caching layer for Database
- Memcached and Redis flavors
- DynamoDB Accelerator
- Native caching layer for DynamoDB reads
- AWS Identity and Access Management
- Allows creating IAM users (no permissions by default)
- IAM policy is used to deny/grant permissions
- JSON document describing if user can/cannot invoke API
{
"Version": "2012-10-17",
"Statement": {
"Effect": "Allow",
"Action": "s3:ListBucket",
"Resource": "arn:aws:s3:::coffee_shop_reports"
}
}- IAM user groups to give permissions to users
- Root users create users
- Roles for temporary access (similar to without any username/password)
- Central location to manage multiple AWS accounts
- Centralized management for all AWS accounts
- Consolidated billing
- Consolidated billing
- Heirarchical account billing
- Access control to services and APIs using Server Control Policies applied to individual member accounts and Organizational Unit (OU)
- Access and review compliance reports done by
$3^{rd}$ parties - Provides access to agreements and reports
- Check all compliance reports in one place
UDP Flood: Prevented by security groups allowing proper traffic at AWS network level Slowloris attack: Prevented by ELB waiting until request is completed
AWS Shield and AWS WAF for advanced security against DDoS attacks
DynamoDB employs encryption at rest and is integrated with AWS Key Management Service (AWS KMS)
- Automated security assessment tool to improve security and compliance
- Analyze continuous stream of metadata and uses integrated threat intelligence like Machine Learning, IP Address...
- Independant of AWS services
Observing systems, collecting metrics, then using data to make decisions
- Monitor AWS infrastructure and apps in realtime by tracking and monitoring metrics (variables tied to resources)
- Set threshold and trigger alert
- Integrated with SNS
- List metrics in realtime
- Access all metrics from central location
- Gain visibility into apps, infrastructure and services
- Improve Mean Time to Resolution (MTTR) and Total Cost of Ownership (TCO)
- Comprehensive API auditing tool
- Every request to AWS gets logged in CloudTrail engine
- Logs who made the request, which operator, whenm where, IP, response, change, new state, if request was denied
- Evaluate resources against 5 pillars
- Cost optimization
- Performance
- Security
- Fault Tolerance
- Service limits
| Symbol | Explanation |
|---|---|
| âś… | Number of items detected for no problem |
| Number of items recommended investigations | |
| âť— | Number of recommended actions |
Over 40 concerns around mentioned pillars
- Always free
- Lambda free for 1 million invocations
- 12 months free
- S3 free for 12 months for upto 5GB
- Trials
- Lightsail offers 1 month trial of upto 750 hours usage
Services including free tier:
- SageMaker
- Comprehend Medical
- DynamoDB
- SNS
- Cognito
- (More)
AWS Pricing is based on pay as you go. Pay less by reserving and volume based discounts.
- Explore AWS services and create cost estimate for use case
- Cost estimate can be saved and shared
- Pay bills
- Monitor usage
- Analyze cost
- Compare current month usage with previous month
- View month to date spend by service
- View free tier usage by service
- Cost Explorer and budget creation
- Purchase and manage savings plan
- Publish cost and usage reports
- Combined bill for all AWS account usage and bill the organization instead of account
- Bulk usage discount for combined use
- Reserved resources can be shared across organization
- Simplifies billing
- Share savings
- Free feature
- Set custom budgets for scenarios like cost and usage
- Get proactively notified when exceeding budget
- Found at
Billing>Budget>Create Budget
- Visually see and analyze how you are spending money on AWS
- Which services are spending most on
- 12 months historic data
- Grouping by tag is also supported
- Free for all customers
- 24/7 customer support service
- Documentation
- Whitepapers
- Forums
- AWS Trusted Advisor
- AWS Personal Health Dashboard
- Basic support
- Email access to customer support
- 12 hour response time
- Basic Support + Developer Support
- Full set of best practice checks in Trusted Advisor
- Phone access to cloud support engineer with 4 hours SLA
- Infrastructure event management for planning massive events
- Basic Support + Developer Support + Business Support
- 15 minute SLA for business critical workloads
- Technical Account Manager (TAM) is provided
- TAMs work with customers to review architectures using Well-Architecture Framework
- 5 pillars of WAF
- Operational Excellence
- Security
- Reliability
- Performance Efficiency
- Cost Optimization
- Digital catalog to find, manage and deploy
$3^{rd}$ party software - One click deployment
- Pay as you go
- Free tier
- Quickstart
Enterprise focussed features - Custom terms and pricing
- Private marketplace
- Cost management tools
- Integration into procurement systems
- Provide advice to your company for quick and smooth migration to AWS
- Perspectives
- Business: Business or finance analyst
- Ensures IT aligns with business needs
- People: HR
- Develop organization-wide change management strategy
- Governance: Program Managers
- Skills and processes to align IT strategy with business strategy
- Platform: Cloud architect
- Principles and platforms for implementing new solutions on cloud
- Security:
- Ensures organization meets security objectives for visibility, auditability, control, agility
- Operations:
- Run, use, operate, recover IT workloads to agreed levels
- Business: Business or finance analyst
- CAF Action Plan helps guide your organization for cloud migration
- Platform perspective helps design, implement, optimize AWS infrastructure based on business goals
6 Options for migration
- Rehosting/Lift and Shift: No changes to app, just pick up the app and move to AWS
- Replatforming/Lift, tinker and shift: Lift and shift with few cloud adoptions and no code change
- Retire: Retire apps which are not required
- Retain: Apps which are going to be deprecated can deprecate where they live without moving
- Repurchase: Abandon legacy software and get a fresh start, example: db vendor to cloud native DB
- Refactoring: Change to architecture
- Designed to enable architects, developers and users to build secure, high performing resilient and efficient infrastructure
- Operational excellence: Running and monitoring systems to deliver business value and improving process and procedures
- Security: Checking data integrity
- Reliability: Recovery planning
- Performance efficiency: IT and computing resource efficiency
- Cost optimization: Optimizing full cost
- Accessed via AWS Management console
- Customizable
-
- Named workloads
- Pillars and dropdowns
- Questions
- Pillars and question stem
- Background and recommendation
- Answer provided and applicability
- Resources
- 2 CPUs
4 GB memory
8 TB storage - 8 TB of data and contain edge computing (EC2 instances and IoT Greengrass)
- Order via AWS management console
Copy your data and ship back
Data then copied to AWS account usually on S3 buckets - Newest offering
- Two versions
- Compute optimized
42 TB storage
52 vCPU
20 GiB memory - Storage optimized
80 TB
40 vCPU
80 GiB
- Compute optimized
- Can fit into existing server racks and clustured
- Once plugged into your infra, can run lambda, EC2 compatible AMIs and IoT Greengrass
- 100 Petabytes
- Ideal for largest migrations and data center shutdowns
-
- Tamper resistant
- Waterproof
- Temperature controlled
- Fire supression
- GPS tracking
- 24/7 video surveillance
- Dedicatged security team
- Pulled by tracks, 45 foot long shipping container
On-premise VMWare instance dropped into AWS
Quickly build, train, deploy ML models at scale or build custom models
ML platform business for building AI
- Ready-to-go AI solutions
- Heart of Alexa
- Helps building chat bots
Extracting text and data from documents
Developers to experiment reinforcement learning
Pay as you go satellite service
flowchart RL
ec2((EC2))
elb((ELB))
sqs((SQS))
sns((SNS))
lambda((Lambda))
ecs((ECS))
eks((EKS))
fargate((Fargate))
bracket((Bracket))
cloudfront((Cloudfront))
route53((Route 53))
outposts((Outposts))
beanstalk((Elastic Beanstalk))
cloudformation((Cloud Formation))
ebs((EBS))
s3((S3))
efs((EFS))
rds((RDS))
aurora((Aurora))
dynamodb((DynamoDB))
redshift((Redshift))
dms((DMS))
amazonmanagedblockchain((Amazon Managed Blockchain))
elasticache((Elasticache))
dax((DAX))
organizations((Organizations))
awsartifact((AWS Artifact))
awscompliance((AWS Compliance))
awsshield((AWS Shield))
waf((WAF))
inspector((Inspector))
guardduty((GuardDuty))
cloudwatch((Cloudwatch))
cloudwatchalarm((Cloudwatch Alarm))
cloudwatchdashboard((Cloudwatch Dashboard))
cloudtrail((CloudTrail))
trustedadvisor((Trusted Advisor))
bacmd((Billing and Cost Management Dashboard))
consolidatedbilling((Consolidated Billing))
budgets((Budgets))
costexplorer((Cost Explorer))
caf((CAF))
wat((WAT))
sagemaker((Sagemaker))
a2i((A2I))
lex((Amazon Lex))
textract((Amazon Textract))
deepracer((DeerRacer))
groundstation((Ground Station))
compute[Compute]
container[Container]
serverless[Serverless]
networking[Networking]
storage[Storage]
database[Database]
blockchain[Blockchain]
iam[IAM]
compliance[Compliance]
security[Security]
monitoring[Monitoring]
pricing[Pricing]
migration[Migration]
mlai[ML/AI]
satellite[Satellite]
ec2 --> compute
elb --> compute
lambda --> compute
lambda --> container
lambda --> serverless
ecs --> compute
ecs --> container
eks --> compute
eks --> container
fargate --> compute
fargate --> container
fargate --> serverless
bracket --> compute
cloudfront --> networking
outposts --> networking
route53 --> networking
outposts --> networking
beanstalk --> serverless
cloudformation --> compute
cloudformation --> database
ebs --> compute
ebs --> storage
s3 --> storage
efs --> storage
rds --> database
aurora --> database
dynamodb --> database
redshift --> database
redshift --> storage
dms --> database
dms --> storage
amazonmanagedblockchain --> blockchain
amazonmanagedblockchain --> storage
elasticache --> database
dax --> database
organizations --> iam
awsartifact --> compliance
awscompliance --> compliance
awsshield --> security
inspector --> security
guardduty --> security
cloudwatch --> monitoring
cloudwatchalarm --> monitoring
cloudwatchdashboard --> monitoring
cloudtrail --> monitoring
trustedadvisor --> monitoring
bacmd --> pricing
consolidatedbilling --> pricing
budgets --> pricing
costexplorer --> pricing
caf --> migration
waf --> security
waf --> migration
wat --> migration
sagemaker --> mlai
a2i --> mlai
lex --> mlai
textract --> mlai
deepracer --> mlai
groundstation --> satellite
- Regions made of multiple data centres
- Regions are geographically isolated
- Each availability zone is one or more data centre within a region with redundant power and networking
- Multiple isolated regions within a geographic region
- Best practice to run atleast 2 availability zones
flowchart
region[REGION]
az1[AVAILABILITY ZONE]
az2[AVAILABILITY ZONE]
dc1[DATA CENTER]
dc2[DATA CENTER]
region --> az1
region --> az2
az1 --> dc1
az1 --> dc2