Bill Shocks, Migration, Exit and Official Sources
Cloud bill shock: 12 common ways Indian small businesses overspend on hosting, and how to avoid them
Twelve generic ways Indian small businesses overspend on hosting and cloud, from idle servers and snapshots to egress and renewal jumps.
Most surprise hosting bills come from a short list of causes: resources left running, storage that keeps growing, data transfer nobody estimated, renewal prices far above the first-term offer, dollar bills that move with the exchange rate, and commitments that outlive the workload. None needs advanced tools to prevent; they need an owner, a budget alert and a monthly review. This guide is the overview for the section. The patterns are generic, drawn from how hosting is billed, not from named companies. Each links to a detailed guide.
The twelve patterns
| # | Pattern | Why it happens | Fix |
|---|---|---|---|
| 1 | Idle servers left running | Test or project servers forgotten after use | Tags with owner and end date; weekly idle report |
| 2 | Oversized servers | Sized on a guess or for a peak | Rightsize after 2–4 weeks of metrics |
| 3 | Orphaned disks and IPs | Deleting a VM leaves disks and IPs billing | Monthly clean-up; delete-on-terminate settings |
| 4 | Snapshot pile-up | No retention policy | Lifecycle rules; see backups and DR |
| 5 | Egress surprise | Images, downloads, backups leaving the cloud | Estimate first; CDN; see egress |
| 6 | NAT and cross-zone traffic | Architecture choices with per-GB charges | Review the network bill line by line |
| 7 | Renewal jump | Introductory price on shared hosting or VPS | Check renewal price before buying; see renewal jumps |
| 8 | Dollar drift | USD billing and card markup | INR billing or FX buffer; see USD vs INR |
| 9 | Free credits ending | Startup or trial credits expire | Diary the expiry; forecast the bill without credits |
| 10 | Wrong commitment | Committing before measuring or before a migration | Commit only the baseline; see break-even |
| 11 | Support plan creep | Percentage-of-bill support tiers | Review whether you need the tier |
| 12 | Leaked credentials | Attackers run servers for crypto-mining at your cost | MFA, no keys in code, budget alerts |
Pattern 1–3: paying for what nobody uses
A developer spins up three servers to test a feature and moves on. A month later they are still running. Or a VM is deleted but its disk and a reserved IP remain. Each item is small, together they are not.
Illustrative. Three forgotten test VMs at ₹3,000 each, two 100 GB orphan disks at ₹800 each and three idle IPs at ₹320 each: ₹9,000 + ₹1,600 + ₹960 = ₹11,560 a month, or ₹1,38,720 a year before GST.
Fix: require tags (owner, project, end date) on every resource, run a weekly report of untagged or idle resources, and switch off non-production servers outside office hours. See budgets, alerts and tagging.
Pattern 4–6: storage and network that grow quietly
Snapshots taken daily without deletion double storage every few months. Egress grows with traffic and is billed after the fact. A NAT gateway processing software updates for twenty servers may cost more than the servers. Read the bill by service, not just the total.
Pattern 7–9: price changes you agreed to without noticing
- Renewal: shared hosting and VPS offers often apply only to the first term.
- Currency: a dollar bill paid by card is converted at the card rate plus a markup on the charge date.
- Credits: free or promotional credits hide the real bill until they expire.
Use the renewal uplift calculator and the GST and forex calculator before the renewal or credit-expiry date.
Pattern 10–11: commitments and plans that no longer fit
A three-year commitment bought before a migration to managed services can leave you paying for servers you no longer run. Support tiers priced as a percentage of the bill grow with every increase in usage.
Pattern 12: someone else's bill
Leaked access keys or a compromised server can be used to launch many large servers. The bill can arrive before you notice. Use multi-factor authentication, keep keys out of code repositories, restrict regions you do not use and set a budget alert at a low level. If you detect such an incident, check whether it must be reported to CERT-In within six hours under its 2022 directions.
Migration and exit: the one-time shocks
Two one-time costs are often missing from plans:
- Migration: parallel running of old and new systems, data transfer and consultant time. See migration cost and payback.
- Exit: data transfer out, rebuilding on the new provider and notice periods. See exit costs and lock-in.
A monthly 30-minute review
- Compare this month's bill with last month by service.
- Check the top five resources by cost.
- List untagged and idle resources; delete or switch off.
- Check snapshot and backup storage growth.
- Check data transfer and NAT lines.
- Note renewal dates, credit expiry and commitment expiry for the next 90 days.
- Record the exchange rate used if billed in dollars.
The official sources guide lists where to check tax and regulatory rules.
FAQ
Can I get a refund for a bill shock?
Sometimes providers give a one-time credit for genuine mistakes, but there is no right to it. Contact support quickly with details.
What budget alert levels should I set?
Common practice is alerts at 50%, 80% and 100% of the monthly budget, plus a forecast alert.
Does shared hosting have bill shocks?
Less often, but renewal prices, add-ons and overage for resources can surprise.
Who should own cloud cost?
A named person, often the IT lead with a finance reviewer. Without an owner, nobody acts on alerts.
Sources
- CERT-In — incident reporting within 6 hours
- CBIC GST — GST on hosting and reverse charge
- Reserve Bank of India — card and e-mandate rules for recurring international payments
- MeitY — DPDP Act and Rules
