Day 1: Why move the shop?

Renting the kennel instead of building one

Cloud computing is the on-demand delivery of IT resources (servers, storage, databases, software) over the internet with pay-as-you-go pricing. Kibble & Co. owns its closet server like a family that built its own dog kennel. In the cloud it rents space in a giant, professionally run kennel: Amazon Web Services (AWS) owns the buildings, power and hardware, and Kibble & Co. pays only for what it uses, for as long as it uses it.

Ranger: sniffs the closet server, sneezes at the dust

Exam tip Exam wording to spot: "on demand", "pay as you go", "no up-front commitment". Those phrases describe the cloud's value proposition.

Six reasons companies move

AWS sums up the value proposition of the AWS Cloud in six advantages. Learn them as a set: the exam often describes one in plain words and asks which benefit it is.

• Trade fixed expense for variable expense: pay for what you use, not for servers you might need
• Benefit from massive economies of scale: AWS buys in bulk, so prices can be lower
• Stop guessing capacity: add or remove resources as demand changes
• Increase speed and agility: new resources in minutes, not weeks
• Stop spending money running and maintaining data centers: focus on customers instead
• Go global in minutes: deploy close to users around the world

Scalability, elasticity and agility

Scalability is the ability to grow to handle more work: bigger servers or more of them. Elasticity goes further: capacity grows and shrinks automatically to match demand, minute by minute, so the shop has extra servers on Black Friday and gives them back on Monday. Agility means you can try an idea quickly and cheaply, and throw it away if it fails, because nothing was bought up front.

Exam tip Trap: "elastic" means automatic AND in both directions (out and back in). A system that can only be upgraded by hand is scalable, not elastic.

High availability

High availability means a system keeps working when a part of it fails, because it runs in more than one place. The closet server is a single point of failure: if its power supply dies, the shop is down. In AWS, the same app can run in two or more separate data centers in the same area. One fails, the other keeps taking orders.

Dev: We'll see exactly how AWS arranges those data centers later. For now: two of everything beats one of anything.

Speed of deployment and global reach

AWS global infrastructure is a worldwide network of data centers that customers can use without building anything. That gives two benefits the exam names: speed of deployment (a test server in minutes instead of a six-week hardware order) and global reach (Kibble & Co. could serve a new Canadian or European store from nearby AWS locations without opening an office there).

Exam tip If a question says "a new market on another continent, launched next week", think global reach. "A test environment today, deleted tomorrow" is speed and agility.

Fixed costs vs variable costs

A fixed cost stays the same no matter how much you use: the server you bought costs the same whether the shop sells ten bags of kibble or ten thousand. A variable cost rises and falls with use. Buying hardware up front is capital expenditure (CapEx); paying as you go is operational expenditure (OpEx). Moving to AWS swaps much of the fixed CapEx for variable OpEx.

Exam tip Exam pattern: "trade fixed expense for variable expense" is the cloud benefit when a question mentions avoiding large up-front purchases.

What the closet really costs

On-premises means running your own equipment in your own building. Its costs go far beyond the server's price tag. Adding them all up over the years gives the total cost of ownership (TCO).

• Hardware bought up front, and replaced every few years
• Power for the server and cooling for the room
• Floor space, racks and physical security
• Staff time for setup, patching hardware and repairs
• Software licenses and maintenance contracts
• Spare capacity bought "just in case" that sits idle most of the year

Economies of scale

Economies of scale means a bigger buyer pays less per unit. AWS buys hardware, power and networking for millions of customers, so its cost per server is lower than Kibble & Co. could ever get, and it has lowered prices many times over the years. You share in that by renting instead of buying.

Mr. Barkley: So they buy a million servers and I rent a slice of the discount. I can live with that.

Rightsizing

Rightsizing means matching the size and type of a resource to what the workload really needs, then checking again as things change. On-premises you buy big "just in case". In AWS an oversized server costs extra every hour, so you shrink it. Don't rent a horse trailer to carry one puppy.

Exam tip Rightsizing is a cost-saving step you repeat, not a one-time purchase decision. AWS Compute Optimizer and AWS Trusted Advisor (later chapters) help spot oversized resources.

Licenses: bring your own, or included

Software like Windows Server or a commercial database needs a license. With Bring Your Own License (BYOL) you reuse licenses you already own on AWS, when the license terms allow it. With license included, the license is part of the hourly price and AWS handles it. BYOL can save money if you already paid for licenses; license included is simpler and suits short or changing use. AWS License Manager helps track both.

Exam tip If a question says "the company already owns licenses and wants to reuse them", the answer is BYOL.

Automation is a cost saver

Anything people do by hand costs staff time and invites mistakes. In AWS you can automate: scripts or templates build the same setup every time, scaling happens without a person watching, and test servers can switch off at 6 p.m. by schedule. The benefits of automation: fewer errors, faster changes, repeatable results and staff freed for work that matters.

Dev: Automation is how a team of four runs a shop that never sleeps.

Ranger: yawns, very much asleep

CapEx to OpEx

On-premises you pay up front for servers and data centers (capital expense, CapEx) and guess how much you'll need. In the cloud you pay for what you use (operational expense, OpEx), with no long-term guessing. Fixed costs become variable costs.

Total cost of ownership

Comparing cloud to on-premises fairly means looking at total cost of ownership (TCO): not just servers but power, cooling, floor space, hardware refreshes, licenses and the staff time to rack, patch and replace things. The cloud also brings soft savings: faster time to market, less time on undifferentiated heavy lifting, and the ability to experiment cheaply.

• Direct costs: compute, storage, data transfer out
• Indirect costs avoided: facilities, hardware maintenance, idle capacity
• Operational savings: managed services do patching and backups

Exam tip The exam calls this 'stop guessing capacity' and 'trade fixed expense for variable expense'.

Two ways to pay for software

License included: the software license (for example Windows Server or Microsoft SQL Server on Amazon EC2 or Amazon RDS) is part of the hourly price; AWS handles it. Bring Your Own License (BYOL): you use licenses you already own, so you pay only for the infrastructure.

Choosing and tracking

BYOL can save money when you already have licenses (some require Dedicated Hosts so you can count physical sockets or cores). License included is simpler and pay-as-you-go. AWS License Manager tracks license usage and rules across accounts and on-premises so you stay compliant.

Exam tip A question about 'existing per-socket licenses' usually points to BYOL on Dedicated Hosts.

What it means

Rightsizing is matching instance types and sizes to what a workload actually uses. Servers are often bought too big 'just in case'; in AWS you can measure usage and move to a smaller (or bigger) size at any time.

• Look at CPU, memory, network and disk usage over time in Amazon CloudWatch
• Get suggestions from AWS Compute Optimizer and AWS Cost Explorer rightsizing recommendations
• Downsize, change instance family, or switch off idle instances
• Repeat regularly: rightsizing is ongoing

Why it's first

Rightsize before buying Savings Plans or Reserved Instances, otherwise you lock in a discount on capacity you don't need.

Exam tip Underused instance + lower bill wanted = rightsizing. Steady 24/7 usage + lower bill wanted = Savings Plans / Reserved Instances.

What automation gives you

In AWS almost everything can be created and changed through APIs, so it can be automated.

• Consistency: the same template builds the same environment every time (fewer human mistakes)
• Speed: new environments in minutes
• Scalability: Auto Scaling adds and removes capacity as demand changes
• Lower cost: switch things off automatically when not needed
• Repeatable recovery: rebuild an environment from code after a failure

Tools you'll see

AWS CloudFormation (infrastructure as code templates), Amazon EC2 Auto Scaling, AWS Systems Manager (patching and run commands across fleets), AWS Elastic Beanstalk (deploys your app and its infrastructure for you).

Big buyer, lower prices

AWS runs hardware for millions of customers, so it buys and operates it at a lower cost per unit than any one company could. Those savings show up as lower pay-as-you-go prices, and AWS has cut prices many times as it has grown.

Exam tip One of AWS's 'six advantages of cloud computing': benefit from massive economies of scale.

The six advantages

AWS's list, often tested as a set:

• 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