Moving to AWS: a business owner's guide

If you’re evaluating a cloud migration and haven’t picked a provider, AWS is the default candidate, and defaults deserve scrutiny. Here’s what AWS actually is, what it costs, and when it’s the right call versus Azure or Google Cloud.

What AWS is, in one minute

Amazon Web Services is the largest cloud provider, holding roughly 30% of the global market (Azure ~25%, Google Cloud ~11-13%, per Synergy/Canalys estimates). It rents you computing infrastructure, servers, databases, storage, networking, by the hour or the second, which is the IaaS model, plus a few hundred higher-level services on top.

Three things follow from being the biggest:

Depth. Over 200 services. Whatever odd requirement you have, GPU fleets, satellite ground stations, a queue that does exactly one thing, AWS probably has a managed version.
Ecosystem. The largest pool of engineers, consultants, documentation, Stack Overflow answers, and third-party tooling. Hiring “someone who knows AWS” is easier than for any other cloud.
Maturity. Core services like EC2 (2006) and S3 (2006) are two decades old. The failure modes are documented; the sharp edges are known.

The honest flip side: 200+ services means a confusing console, a real learning curve, and a pricing page that requires effort to read. AWS is not the simplest cloud, it’s the most complete one.

The five services you’ll actually use

Most migrations, whatever the marketing says, touch a handful of primitives. Learn these five and you understand 80% of a typical AWS bill:

Service	What it is	On-prem equivalent	Rough cost anchor
EC2	Virtual servers you size and manage	Your VMware VMs / physical servers	~$30/mo for a 2 vCPU / 4 GB (t3.medium) on-demand
RDS	Managed databases (PostgreSQL, MySQL, SQL Server), AWS handles patching, backups, failover	Your database server + the DBA time	~$60-120/mo for a small instance; ~2× with a standby replica
S3	Object storage, files, backups, static assets, effectively unlimited	File servers, NAS, tape backups	~$0.023/GB-month, cheaper tiers for archives
VPC	Your private network in AWS, subnets, routing, firewalls	Your LAN, VLANs, and firewall rules	Free itself; NAT gateways ~$32/mo + data charges
IAM	Who and what can do which actions on which resources	Active Directory permissions, but for infrastructure	Free, and the thing most often misconfigured

A VPC and IAM setup done properly on day one is the difference between a secure account and a future incident, that’s the core of a landing zone.

Beyond the five: ELB (load balancers), CloudWatch (monitoring), Lambda (serverless functions), and ECS/EKS (containers) show up in most real deployments. Ignore the rest until you need it.

How AWS pricing actually behaves

The model is pay-per-use: per-second compute billing, per-GB storage, per-GB data transfer. That’s genuinely good for variable workloads and genuinely dangerous for the unprepared. The gotchas, in order of how often we see them:

Data egress. Data in is free; data out to the internet costs ~$0.09/GB after the first 100 GB/month. Serving 5 TB of downloads or video a month is ~$440, a line item on-prem never had. Egress is also the quiet lock-in: moving 50 TB out of AWS later costs real money.
Paying on-demand for steady workloads. On-demand is the rack rate. A server you know will run all year should be on a Savings Plan or Reserved Instance, typically 30-45% cheaper for a 1-year commitment, up to ~60-70% for 3 years. Most new accounts leave this money on the table for months.
Forgotten resources. Unattached disk volumes, idle NAT gateways, old snapshots, oversized instances “to be safe.” Cost reviews of 1-2-year-old accounts routinely find 20-35% waste. Tag everything and review monthly, more in how to avoid bill shock.
Cross-AZ and NAT traffic. Traffic between availability zones and through NAT gateways is metered (~$0.01-0.045/GB). Chatty architectures generate surprising network line items.

Worked example, a typical small SaaS or internal platform:

Item	Spec	On-demand/mo
2 × app servers	t3.medium	$60
Database	RDS PostgreSQL db.t3.medium, Multi-AZ	$125
Storage	250 GB S3 + 200 GB volumes	$26
Load balancer	ALB	$25
NAT gateway	1	$35
Egress	500 GB/mo	$45
Total		≈ $316/mo

With a 1-year Savings Plan on the compute, closer to $240/month. Estimate yours before you move with our cost estimation guide, and note the same stack prices within ±15% on Azure or Google Cloud; the pricing comparison matters less than your discipline.

When AWS is the right pick, and when it isn’t

AWS is the strong default when:

Your team is mixed-stack (Linux, open-source databases, containers) rather than Microsoft-everything.
You need hiring depth, the AWS talent pool is the largest by a wide margin.
You want the broadest service catalog so you won’t outgrow the platform in year three.
You’re building customer-facing products where the ecosystem of reference architectures and third-party integrations pays off daily.

Look elsewhere first when:

You’re a Microsoft shop. Windows Server, SQL Server, Active Directory, Microsoft 365 everywhere, Azure’s Hybrid Benefit licensing and native Entra ID integration typically save 20-40% on those workloads and a lot of identity plumbing.
Data and ML are the product. Teams that live in analytics pipelines and Kubernetes often move faster on Google Cloud, BigQuery and GKE remain best-in-class.
Your workload is one boring server. A predictable app with a few hundred users runs fine on a $20-60/month VPS. The cloud’s value is elasticity, managed services, and reliability engineering, if you need none of those yet, don’t pay the premium.

Vendor-neutral truth: all three hyperscalers are reliable, secure, and roughly price-comparable for standard workloads. The pick is about your team’s skills, your existing licensing, and your roadmap, not about one cloud being “better.”

How a migration to AWS actually goes

The shape is consistent whether it’s 1 application or 40. Full detail in Migrating from on-premise to AWS and the migration roadmap; the summary:

Discovery (1-2 weeks). Inventory applications, servers, databases, and the dependencies between them. Decide each workload’s path, rehost, replatform, retire, per the 6 Rs.
Landing zone (about a week). Accounts, VPC networking, IAM, logging, and guardrails before the first workload lands. Skipping this is the most expensive shortcut in cloud migration.
Replicate and test (2-4 weeks per wave). Stand up infrastructure, replicate data continuously (e.g., DMS for databases), and test against production-like data while the old system keeps running.
Cutover (minutes to hours). DNS switch during a low-traffic window, with a tested rollback path. Done right, users notice nothing, see zero-downtime migration.
Optimize (ongoing). Right-size, buy Savings Plans, kill waste. The first bill is not your real bill; the 90-day-tuned bill is.

Realistic timeline: a single app in 4-8 weeks; a 10-20 app portfolio in 4-9 months. Anyone quoting “two weeks for everything” hasn’t done discovery yet.

FAQ

Is AWS good for small businesses?

Yes, with a caveat. AWS scales down fine, a small workload can run for $50-300/month, but the console and pricing model assume some engineering literacy. If nobody on your team has cloud experience, budget for a partner or a managed service, or you’ll overpay and misconfigure security.

How much does it cost to run a typical business application on AWS?

A common small-business stack, two application servers, a managed PostgreSQL database with a standby, storage, and a load balancer, lands around $250-500/month on-demand, and roughly 30-40% less with a savings plan. Costs scale mostly with database size and data transfer out.

How long does a migration to AWS take?

A single application with one database typically takes 4-8 weeks end to end: 1-2 weeks discovery and landing-zone setup, 2-4 weeks replication and testing, then a cutover window measured in minutes to hours. A portfolio of 10-20 applications is more like 4-9 months.

When is AWS the wrong choice?

When your company runs on Microsoft everywhere (Azure’s licensing and Entra ID integration will save real money), when your differentiator is data/ML and your team lives in BigQuery-style tooling (GCP), or when your workload is one predictable server that a $40/month VPS handles fine.

Planning a move to AWS and want a second pair of eyes on the architecture, the cost estimate, or the cutover plan? Talk to a Webisoft cloud engineer, we’ll review your setup and give you a straight answer, including “you don’t need us for this” when that’s true.