Chapter 75: AWS Pricing and Support
AWS Pricing and Support
This is not just “how much does AWS cost?” It is a complete system — how AWS charges you, how you can control & predict your bill, what free tiers exist, what support plans are available, and how to avoid the classic ₹50,000–₹2 lakh surprise bills that many Hyderabad startups face in their first 3–6 months.
I will explain it like we’re sitting together with a whiteboard and a second cup of filter coffee — slow, clear, step-by-step, full of real analogies, actual 2026 Hyderabad startup examples, current pricing reality in ap-south-1 & ap-south-2, and exactly how smart teams keep their bills predictable and support useful.
1. The Golden Rule of AWS Pricing (Memorize This First)
AWS Pricing Principle #1 (2026 reality):
You pay only for what you use — but “use” includes everything you forget to turn off.
Unlike traditional servers where you buy a fixed machine and pay the same every month, AWS is pay-as-you-go — but that means idle resources, forgotten test environments, and misconfigured auto-scaling can cost you much more than a fixed server ever would.
The most common bill shocks in Hyderabad startups in 2026 come from exactly these:
- Left EC2 instances running in us-east-1 (very expensive region)
- Public S3 buckets with thousands of GET requests from bots
- Forgot to delete EMR/SageMaker training clusters
- Lambda functions with recursive loops
- RDS Multi-AZ + Provisioned IOPS way higher than needed
2. The Main Pricing Models on AWS (How You Actually Get Billed)
| Pricing Model | How it works (simple) | Typical services using this model | Best for (Hyderabad startup example) |
|---|---|---|---|
| On-Demand | Pay by the second / hour — no commitment | EC2, RDS, Lambda (default), ECS Fargate | Dev/test, unpredictable traffic |
| Savings Plans | Commit to 1 or 3 years of usage — up to 72 % discount | EC2, Fargate, Lambda, SageMaker | Production workloads with steady usage |
| Reserved Instances (RI) | Commit to specific instance type & term (1 or 3 years) | EC2, RDS, Redshift | Predictable long-running instances |
| Spot Instances | Bid on spare capacity — up to 90 % cheaper, can be interrupted | EC2 Spot, ECS Spot tasks, Batch jobs | Non-critical batch jobs, ML training |
| Pay-per-request / Serverless | Pay only for actual usage (requests, duration, GB-seconds) | Lambda, DynamoDB (on-demand), Athena, Glue, Step Functions | Variable / bursty workloads |
| Data Transfer | Charges for moving data out to internet / between regions | S3, CloudFront, EC2 → internet | Minimize egress costs with CloudFront |
2026 quick rule for Hyderabad startups:
- Start with On-Demand + Free Tier → safe for learning
- Move to Savings Plans as soon as monthly bill > ₹30,000–50,000 (usually covers 70–80 % of usage)
- Use Spot for non-critical batch jobs (CI/CD, ML training)
- Never run production in expensive regions (us-east-1, eu-west-1) — stick to ap-south-1 or ap-south-2
3. The Free Tier — Your Best Friend in the First Year
AWS Free Tier (still very generous in 2026):
| Service | Free Tier Amount (every month, forever unless noted) | Typical Hyderabad startup usage |
|---|---|---|
| EC2 | 750 hours t3.micro / t4g.micro | Dev/test instances running 24/7 |
| S3 | 5 GB Standard storage, 20,000 GET, 2,000 PUT | Small static website + user uploads |
| Lambda | 1 million requests + 400,000 GB-seconds | Background jobs, webhooks |
| DynamoDB | 25 GB storage, 25 write/read capacity units | Small metadata / cart tables |
| RDS | 750 hours db.t4g.micro (Single-AZ) | Dev/test database |
| CloudWatch | Basic monitoring, 10 custom metrics, 1 million API requests | Most dashboards & alarms |
Real example: A college project team in CBIT / Vasavi launches a Node.js API + React frontend + DynamoDB → Runs on t4g.micro (free) + S3 static hosting (free) + Lambda (free) + DynamoDB (free) → Monthly AWS bill = ₹0 for first 12 months — even with 1,000 daily users
4. AWS Support Plans — Which One Should You Choose?
| Plan | Monthly Cost (rough, 2026) | Key Benefits (what you actually get) | Who uses this plan in Hyderabad? |
|---|---|---|---|
| Basic | Free | 24×7 billing & account support, Trusted Advisor (7 checks), forums | Students, very early startups |
| Developer | ~₹2,500–7,500 | Email support during business hours, full Trusted Advisor | Small startups, side projects |
| Business | ~₹7,500–35,000 | 24×7 phone/email support, Infrastructure Event Management, full Trusted Advisor | Most scaling startups & mid-size companies |
| Enterprise On-Ramp | ~₹35,000+ | TAM (Technical Account Manager), Well-Architected reviews, concierge support | Serious fintech, health-tech |
| Enterprise | ₹1,00,000+ | Dedicated TAM, 24×7 priority support, organizational workshops | Large enterprises, banks |
2026 recommendation for Hyderabad teams:
- Month 1–6 → Basic or Developer (free / low cost)
- Month 6–18 → Business Support — when monthly bill > ₹50,000–80,000 → Full Trusted Advisor alone usually pays for the plan in 1–2 months
- After ₹5–10 lakh/month → consider Enterprise On-Ramp for TAM help
5. Real Hyderabad Example — How a Startup Uses Pricing & Support
Your startup “TeluguBites” (restaurant discovery + food ordering) — 8 developers, 3 DevOps, 5 AWS accounts
Pricing journey (typical 2026):
- Month 1–3 — Basic support + Free Tier → Bill ~₹8,000–15,000 (mostly EC2 On-Demand + RDS) → Used Cost Explorer to see “dev account” spending 60 % on idle t3.medium instances → stopped them
- Month 4 — Bill crossed ₹45,000 → Upgraded to Business Support (~₹12,000/month) → Full Trusted Advisor found:
- 4 idle EC2 instances → stopped → saved ₹9,000/month
- RDS Single-AZ → enabled Multi-AZ → added high availability
- Security group allowing 0.0.0.0/0 on port 22 → fixed in 5 min
- Month 8 — Bill ~₹1.2 lakh/month → Bought Compute Savings Plan (1-year, all-in) → 65 % discount on ECS Fargate & Lambda → Bill dropped to ~₹70,000/month for same usage
- Support usage:
- Called 24×7 support twice: once for RDS Multi-AZ failover question, once for VPC peering limit increase
- Saved ~3 days of debugging
Summary Table — AWS Pricing & Support Cheat Sheet (2026 – India Focus)
| Question | Answer (Beginner-Friendly) |
|---|---|
| How does AWS charge? | Pay-as-you-go — seconds/hours/requests/GB — you pay for what you use (and what you forget to turn off) |
| Free Tier — how long? | Most services “forever free” tier (e.g., 750 h EC2 t4g.micro every month) |
| When to upgrade support? | When monthly bill > ₹50,000–80,000 — Business Support usually pays for itself |
| Best way to control costs? | Tag everything, use Cost Explorer + Budgets alerts, buy Savings Plans early |
| Biggest bill shock causes? | Idle EC2/RDS, public S3 GETs from bots, non-approved regions (us-east-1) |
| First thing to do today? | Open Cost Explorer → see last 30 days usage → stop any idle t3/t4g instances |
Teacher’s final note (real talk – Hyderabad 2026):
AWS pricing is very fair — but it punishes forgetfulness and rewards discipline.
The teams that keep bills low & predictable in Hyderabad are not the ones using the cheapest services — they are the ones who:
- Tag every resource (Environment, Owner, Project)
- Check Cost Explorer weekly
- Set Budgets alerts at 80 %
- Buy Savings Plans as soon as usage is predictable
- Use Business Support for Trusted Advisor once bill > ₹50,000
Do these five things religiously and your AWS bill will stay predictable & reasonable even as you scale.
Got it? This is the “how to use AWS without getting a heart-attack bill” lesson.
Next?
- Step-by-step: Set up Cost Explorer + Budgets + tag policy in a new account?
- Deep dive: Savings Plans vs Reserved Instances — which one saves more in 2026?
- Or how to use Trusted Advisor Cost Optimization checks to find idle resources?
Tell me — next whiteboard ready! 🚀💰
