Solution Architecture overview

80 topics in 13 lessons, each in one line. Every topic opens the page that teaches it.

After this course you can

  • Turn a brief into numbers (requests per second, storage, latency and availability targets) and design to them.
  • Choose and combine load balancers, caches, databases, queues, functions and containers, and say what each choice costs.
  • Design for failure and disaster, secure the network and the data, and review a design against the six Well-Architected pillars.

1Requirements and estimates

2Availability

3Scaling

4Load balancing

5Caching and content delivery

6Databases at scale

7Queues and events

8Monoliths, microservices and serverless

9Designing for failure

10Disaster recovery

11Security and network design

12Cost, operations and reviewing a design

13A photo app from a thousand users to a millionWorked case

Sources and credit

51 sources

The chapters and drawings are written by NeuroqAI from these sources. Each lesson lists its own at the end of its chapter, and figures on the pages are taken from them or computed from them.

Amazon Web Services37 sourcesAWS Well-Architected FrameworkThe six pillars used to review any cloud designLessons 1, 12, 13AWS Certified Solutions Architect Associate (SAA-C03) exam guideThe domains of the associate architect examLessons 1, 12Reliability pillar (AWS Well-Architected)Availability arithmetic, fault isolation and static stabilityLessons 2, 9Amazon Compute Service Level AgreementWhat AWS promises for machines in one zone and across zonesLesson 2Multi-AZ deployments for Amazon RDSA standby in another zone, and how long failover takesLessons 2, 6Amazon EC2 instance typesThe families of virtual machines and what each is sized forLesson 3Amazon EC2 On-Demand pricingMachine prices per hour and data transfer pricesLessons 3, 12What is Amazon EC2 Auto Scaling?Groups of machines that grow and shrink with demandLesson 3What is an Application Load Balancer?Listener rules, target groups and health checksLesson 4What is Elastic Load Balancing?Application, network and gateway load balancersLesson 4Choosing a routing policy (Route 53)Weighted, latency, failover and geographic routingLesson 4Caching strategies (Amazon ElastiCache)Lazy loading, write-through and TTLLessons 5, 13What is Amazon CloudFront?Edge locations, cache keys and originsLessons 5, 13Working with DB instance read replicasAsynchronous copies that take readsLessons 6, 13What is Amazon Aurora?Six copies across three zones under one databaseLesson 6Partitions and data distribution (DynamoDB)Why the partition key decides how far a table scalesLesson 6What is Amazon Simple Queue Service?Queues that let one part of a system wait for anotherLesson 7What is Amazon SNS?One message delivered to many subscribersLesson 7What is Amazon EventBridge?Routing events by ruleLesson 7Understanding Lambda function scalingConcurrency is requests a second times durationLesson 8Lambda quotasThe limits of a function: 15 minutes, memory, payload and concurrencyLesson 8AWS Lambda pricingPrice per request and per GB-second, used for the cost barsLesson 8Exponential backoff and jitter (2015)The simulation behind adding randomness to retriesLesson 9Disaster recovery options in the cloudBackup and restore, pilot light, warm standby and multi-siteLesson 10Using Amazon Aurora Global DatabaseA copy in another region, about a second behindLesson 10Introduction to backups (Amazon RDS)Automated backups kept 1 to 35 days, and point-in-time restoreLesson 10Replicating objects (Amazon S3)Copies of a bucket in another regionLesson 10What is Amazon VPC?Private networks, subnets, route tables and gateways on AWSLesson 11Amazon VPC pricingNAT gateway hourly and per-GB prices used in the examplesLesson 11What is AWS PrivateLink?Reaching services without the internet or a NAT gatewayLesson 11AWS Key Management ServiceKeys that encrypt data, and who may use themLesson 11What is AWS Certificate Manager?Free certificates that renew themselvesLesson 11What is AWS Secrets Manager?Secrets stored, rotated and fetched at run timeLesson 11AWS WAFFiltering web requests before they reach the appLesson 11What is the AWS Well-Architected Tool?Running a review against the six pillarsLesson 12What are Savings Plans?A one- or three-year spend commitment in return for a lower rateLesson 12Amazon S3 storage classesEach S3 class, its retrieval time and what it is forLesson 12
Standards, research and other publishers14 sourcesISO/IEC 25010 quality model · ISO 25000 portalThe qualities a system can be asked to haveLesson 1Documenting architecture decisions (2011) · Michael NygardThe short decision record teams still useLesson 1Latency numbers every programmer should know, by year · Colin ScottJeff Dean's numbers, updated year by yearLesson 1The Tail at Scale (2013) · Jeffrey Dean and Luiz André BarrosoWhy the slowest 1% decides how fast a big system feelsLesson 1Embracing risk (Site Reliability Engineering) · GoogleError budgets: how much failure to allowLesson 2Service level objectives (Site Reliability Engineering) · GoogleSLIs, SLOs and SLAsLesson 2Service Level Agreements for Online Services · MicrosoftAzure's promised availability for each serviceLesson 2What are Virtual Machine Scale Sets? · MicrosoftAzure's groups of identical machines that scale out and inLesson 3RFC 9111: HTTP Caching · IETF (RFC Editor)Cache-Control, max-age and what a shared cache may keepLesson 5Microservices (2014) · James Lewis and Martin FowlerWhat the style is, and what it costsLesson 8Timeouts, retries and backoff with jitter · Amazon Builders' LibraryHow Amazon keeps retries from making outages worseLesson 9Using load shedding to avoid overload · Amazon Builders' LibraryRefuse some work to keep serving the restLesson 9Circuit breaker · Martin FowlerStop calling a service that is failingLesson 9What Powers Instagram: Hundreds of Instances, Dozens of Technologies · Instagram EngineeringThe whole stack at 14 million users, run by three engineersLesson 13