Cloud computing is the on-demand delivery of IT resources over the internet with pay-as-you-go pricing. AWS owns and maintains the hardware; you provision what you need and pay for what you use.
Benefits of the AWS Cloud
Cloud computing delivers IT resources on demand over the internet, with pay-as-you-go pricing. AWS owns and maintains the hardware, and you provision what you need when you need it.
How it works
- You request resources through the console, the CLI or an API instead of buying hardware.
- AWS runs them on its infrastructure in the Region you choose, and bills for what you use.
- When demand changes, you add or remove resources, so capacity follows the workload. This is elasticity.
- Spreading a workload over many small resources and over separate locations keeps one failure from taking it all down. This is high availability.
AWS names six advantages of cloud computing:
| Advantage | What changes |
|---|---|
| Trade fixed expense for variable expense | Pay only when and as much as you consume, instead of buying data centers up front |
| Benefit from massive economies of scale | Aggregated usage lets AWS charge lower pay-as-you-go prices |
| Stop guessing capacity | Scale up and down with minutes of notice instead of buying for the peak |
| Increase speed and agility | Resources in minutes instead of weeks, so experiments cost less |
| Stop spending money running and maintaining data centers | Spend effort on the business, not on racking and powering servers |
| Go global in minutes | Deploy in several Regions for lower latency |
Exam cues
- If the question says avoid a large upfront hardware purchase, think trade fixed expense for variable expense.
- If the question says lower prices because many customers share AWS, think economies of scale.
- If the question says idle servers bought for a peak, or not enough capacity, think stop guessing capacity, because the cloud scales both ways.
- If the question says faster experiments or shorter time to provision, think speed and agility.
- If the question says users in other countries need low latency, think go global in minutes.
- If the question says bigger instance type, think vertical scaling, because it resizes one resource; more instances is horizontal scaling.
Remember
- Six advantages: variable expense, economies of scale, stop guessing capacity, speed and agility, no data center spending, global in minutes.
- Elasticity scales in both directions; you pay for what runs.
- Many small resources with no common point of failure raise availability.
Trap: Economies of scale is about AWS’s lower prices from aggregated usage, not about your own workload growing. A question about a single company’s traffic spikes points to elasticity (stop guessing capacity).
From the AWS docs (11)
- Cloud computing is the on-demand delivery of compute power, databases, storage, applications and other IT resources over the internet with pay-as-you-go pricing.
- A cloud services platform such as AWS owns and maintains the network-connected hardware; customers provision and use what they need.
- Trade fixed expense for variable expense: instead of investing in data centers and servers up front, pay only when you consume computing resources and only for how much you consume.
- Benefit from massive economies of scale: usage from hundreds of thousands of customers is aggregated, so AWS achieves higher economies of scale that translate into lower pay-as-you-go prices.
- Stop guessing capacity: instead of sitting on idle resources or running short, access as much or as little capacity as needed and scale up and down with a few minutes' notice.
- Increase speed and agility: new IT resources are a click away, cutting the time to make them available to developers from weeks to minutes and lowering the cost of experimenting.
- Stop spending money running and maintaining data centers: focus on projects that differentiate the business instead of racking, stacking and powering servers.
- Go global in minutes: deploy an application in multiple Regions around the world with a few clicks, giving customers lower latency at minimal cost.
- Replacing one large resource with multiple small resources, and distributing requests across them so they share no common point of failure, increases aggregate workload availability.
- Monitoring a workload for key performance indicators lets automation run when a threshold is breached, so recovery from failure can be automatic.
- Scaling horizontally (out) replaces one large resource with several smaller ones so a single failure has less impact; scaling vertically (up) makes one resource bigger and adds no redundancy.
Cloud deployment models
A deployment model says where an application’s parts run. AWS describes three: all in the cloud, hybrid, and on-premises. Each gives a different level of control, flexibility and management.
How it works
- Cloud: every part of the application runs in the cloud. It may use low-level infrastructure, such as virtual servers, or higher-level services that take over management and scaling.
- Hybrid: cloud resources connect to resources that stay outside the cloud, usually an existing data center. Companies use it to grow into the cloud while keeping internal systems.
- On-premises (private cloud): virtualization and resource management tools run in your own data center. You get dedicated resources but keep most of the work of legacy IT.
| Model | Where it runs | Typical reason |
|---|---|---|
| Cloud | Entirely on AWS | New applications, or full migrations |
| Hybrid | AWS plus your data center | Systems that must stay put, gradual migration |
| On-premises | Your data center only | Dedicated resources; few cloud benefits |
Exam cues
- If the question says some systems stay in the data center and connect to AWS, think hybrid.
- If the question says the whole application runs on AWS, think cloud.
- If the question says virtualization in the company’s own data center, think on-premises (private cloud).
Remember
- Three models: cloud, hybrid, on-premises.
- Hybrid most often means cloud plus an existing data center.
- On-premises private cloud gives dedicated resources, not the cloud’s benefits.
- IaaS: building blocks, most control. PaaS: provider runs hardware and OS. SaaS: finished software.
Trap: Hybrid does not mean using two cloud providers. AWS uses it for cloud resources connected to resources outside the cloud, most commonly on-premises.
From the AWS docs (6)
- In a cloud deployment, all parts of the application run in the cloud, either built there or migrated from existing infrastructure.
- Cloud-based applications can be built on low-level infrastructure or on higher-level services that abstract away managing, architecting and scaling core infrastructure.
- Deploying resources on premises with virtualization and resource management tools is sometimes called private cloud; it provides dedicated resources but few of the benefits of cloud computing.
- A hybrid deployment connects infrastructure and applications between cloud resources and existing resources outside the cloud, most commonly on-premises infrastructure.
- Each type of cloud computing gives a different level of control, flexibility and management.
- Cloud services are traditionally grouped as IaaS (networking, compute and storage building blocks with the most control), PaaS (the provider manages the underlying hardware and operating systems so you focus on your application) and SaaS (complete software run and managed by the provider).
Cloud economics
Cloud economics compares what a workload costs in a data center with what it costs on AWS. On premises you buy capacity up front and pay for it whether you use it or not; on AWS you pay for what runs.
How it works
- Fixed vs variable costs. A data center needs servers, space, power and staff before the first user arrives. On AWS those become variable costs that follow usage.
- On-premises costs. Beyond hardware, a company pays to rack, stack and power servers and to manage operating systems. AWS takes on the data center work, and managed services take on more.
- Economies of scale. AWS spreads its costs over many customers, so its prices fall below what one company can reach alone.
- Right-sizing. Pick the resource type, size and number that meet requirements at the lowest cost, and revisit it as usage changes.
- Automation. Scripts and services start, stop, resize and recover resources without people, which cuts labor and idle time.
- Licensing. Some software comes license included in the price; with Bring Your Own License (BYOL) you reuse licenses you already own.
| Licensing model | Who holds the license | Example |
|---|---|---|
| License Included | AWS, built into the price | RDS for Oracle Standard Edition 2 |
| Bring Your Own License (BYOL) | You, from licenses you already own | RDS for Oracle Enterprise Edition; Windows Server on EC2 Dedicated Hosts |
Migration Evaluator builds a business case for a move, finds over-provisioned on-premises servers and compares BYOL with License Included.
Exam cues
- If the question says no large upfront investment, think variable expense.
- If the question says servers are bigger than the workload needs, think right-sizing.
- If the question says reuse existing licenses, think BYOL; on EC2 for per-core or per-socket licenses, think Dedicated Hosts.
- If the question says build a business case for migration from on-premises data, think Migration Evaluator.
Remember
- Stopping dev and test outside an 8-hour workday saves about 75% (40 of 168 hours).
- Right-sizing repeats as usage, prices and resource types change.
- RDS for Oracle License Included: Standard Edition 2 only. BYOL: Enterprise Edition and Standard Edition 2.
- Cost drivers: compute, storage, outbound data transfer.
- No extra charge: IAM and EC2 Auto Scaling (you pay for the resources Auto Scaling uses).
Trap: Moving to AWS does not remove every cost of running servers. AWS takes on the data center, but you still choose sizes, stop idle resources and manage what you run on unmanaged services.
From the AWS docs (17)
- In the cloud, a fixed upfront investment in data centers and servers becomes a variable expense paid only for the resources consumed.
- With cloud computing you don't make large upfront investments in hardware or spend time managing that hardware.
- Aggregated usage from many customers gives AWS higher economies of scale, which translate into lower pay-as-you-go prices.
- Adopting a consumption model means paying only for resources consumed; stopping development and test environments outside an 8-hour workday can save about 75% (40 hours instead of 168 per week).
- AWS does the heavy lifting of data center operations such as racking, stacking and powering servers, and managed services remove the burden of managing operating systems and applications.
- The cloud makes it easier to identify the cost and usage of workloads and attribute IT costs to revenue streams and workload owners.
- Right-sizing selects the resource type, size and number that meet technical requirements at the lowest cost, and can repeat as usage, prices or resource types change.
- Resource type, size and number can be selected automatically based on metrics, as well as from data and cost modeling.
- Automation can track failures and run recovery processes when a monitored threshold is breached, and more advanced automation can remediate failures before they occur.
- A common cause of failure in on-premises workloads is resource saturation, when demand exceeds the capacity that was bought.
- Amazon RDS for Oracle offers License Included, where AWS holds the Oracle license, and Bring Your Own License (BYOL), where you use existing Oracle Database licenses.
- In the License Included model you don't buy database licenses separately; RDS for Oracle supports it only for Standard Edition 2.
- EC2 Dedicated Hosts support BYOL for licenses bound to VMs, sockets or physical cores, such as Windows Server and SQL Server, subject to the license terms.
- Migration Evaluator builds a data-driven business case for moving to AWS, finds over-provisioned on-premises instances and compares BYOL with License Included options.
- The three fundamental drivers of AWS cost are compute, storage and outbound data transfer.
- AWS Identity and Access Management (IAM), IAM Identity Center and AWS STS are offered at no additional charge.
- Amazon EC2 Auto Scaling has no additional fee; you pay for the EC2 instances, CloudWatch and other resources it uses.
Flashcards
Nothing to review here right now. Pick other topics or another filter.
·
All 28 cards as a list
What are the six advantages of cloud computing that AWS lists?
- Trade fixed expense for variable expense
- Benefit from massive economies of scale
- Stop guessing capacity
- Increase speed and agility
- Stop spending money running and maintaining data centers
- Go global in minutes
A startup does not want to buy servers before it knows how much its new product will be used. Which advantage of the AWS Cloud addresses this?
Trade fixed expense for variable expense. It pays only when it consumes resources, and only for how much it consumes.
Why can AWS offer lower pay-as-you-go prices than a company running its own data center?
Economies of scale. AWS aggregates usage from hundreds of thousands of customers, which lowers its variable cost.
A retailer buys enough servers for its holiday peak, and they sit idle the rest of the year. Which AWS Cloud advantage removes this problem?
Stop guessing capacity. On AWS it can scale up for the peak and back down with a few minutes' notice.
Developers wait weeks for new servers to test an idea. Which AWS Cloud advantage shortens this to minutes?
Increase speed and agility. New IT resources are a click away, so experimenting costs less time and money.
A company wants its staff to work on products that set it apart instead of racking and powering servers. Which AWS Cloud advantage does this describe?
Stop spending money running and maintaining data centers. AWS does the undifferentiated heavy lifting of the hardware.
A company wants lower latency for customers on three continents without building data centers there. Which AWS Cloud advantage applies?
Go global in minutes. It can deploy the application in Regions near those customers with a few clicks.
How does running a workload on many small resources instead of one large one improve availability on AWS?
A single failure affects a smaller share of the workload, and spreading requests across resources avoids a common point of failure.
Does elasticity in the AWS Cloud mean a workload can only scale up?
No. It means scaling up and down as demand changes, paying only for what is used.
A company moves an application to a bigger EC2 instance type. Which kind of scaling is this, and does it add fault tolerance?
Vertical scaling (scaling up). It adds no fault tolerance, because there is still one instance.
Which kind of scaling adds more EC2 instances behind a load balancer and also improves availability?
Horizontal scaling (scaling out): more, smaller resources, so one failure affects less of the workload.
What are the three cloud deployment models AWS describes?
- Cloud: every part of the application runs in the cloud
- Hybrid: cloud resources connected to resources outside the cloud
- On-premises (private cloud): virtualized resources in your own data center
A company keeps its mainframe in its data center and connects it to new applications running on AWS. Which deployment model is this?
Hybrid. It connects cloud resources to existing resources outside the cloud.
A company moves every component of an application to AWS and shuts down the old servers. Which deployment model is this?
Cloud. All parts of the application run in the cloud.
Does a private cloud built with virtualization in a company's own data center give the same benefits as the AWS Cloud?
No. On-premises private cloud offers dedicated resources but few of the benefits of cloud computing, and it works much like legacy IT.
A team wants to deploy its application code without managing the servers or operating systems underneath. Which cloud service model is this?
Platform as a Service (PaaS).
IaaS gives raw building blocks to manage; SaaS is finished software run by the provider.
Which cloud service model gives the most control over networking, compute and storage?
Infrastructure as a Service (IaaS).
How does moving to the AWS Cloud change the way a company pays for servers?
A fixed upfront investment becomes a variable expense. It pays only for the resources it consumes.
A company stops its development and test environments outside an 8-hour workday, 5 days a week. About how much of their running cost does it save on AWS?
About 75%. They run 40 hours instead of 168 hours a week, and AWS charges only for what runs.
Which on-premises costs does a company stop paying when it moves servers to AWS?
The data center work of racking, stacking and powering servers. With managed services, AWS also takes over operating system and application management.
What does right-sizing mean in AWS cost optimization?
Choosing the resource type, size and number that meet the requirements at the lowest cost, and repeating it as usage or prices change.
A company owns Oracle Database Enterprise Edition licenses and wants to run the database on Amazon RDS. Which licensing model applies?
Bring Your Own License (BYOL). RDS for Oracle offers License Included only for Standard Edition 2.
A company has Windows Server licenses bound to physical cores and wants to use them on Amazon EC2. Which EC2 option supports this?
EC2 Dedicated Hosts. They support BYOL for per-socket, per-core and per-VM licenses.
A company wants a data-driven business case comparing its on-premises costs with AWS, including license options. Which AWS offering helps?
Migration Evaluator. It finds over-provisioned on-premises instances and compares BYOL with License Included costs.
How does automation reduce costs and risk in the AWS Cloud?
Automated actions replace manual work, for example recovering from failure when a monitored threshold is breached, or resizing resources based on metrics.
Does buying the largest servers on premises protect a workload from capacity problems better than the AWS Cloud?
No. Resource saturation is a common on-premises failure, and oversized servers sit idle. On AWS capacity follows demand and is billed as used.
What are the three fundamental drivers of cost on AWS?
Compute, storage and outbound data transfer.
Which AWS services in this list are free to use: IAM, EC2 Auto Scaling, Amazon EC2 instances, Amazon EBS volumes?
IAM and EC2 Auto Scaling.
Auto Scaling itself is free; you pay for the instances and CloudWatch resources it uses.