AWS and Azure Eye India Data Centre Shift: What Indian SMBs Should Check

Your cloud bill probably has a line item you can't explain. Maybe it's data egress charges that jumped after a traffic spike, or a managed database running in Singapore while every customer you have sits in Pune and Chennai. For many Indian SMBs, the question of where a workload runs has quietly become as important as which provider runs it. Recent moves by AWS and Azure to reroute and expand capacity for workloads originating in West Asia toward Indian regions have made that question more pressing, not less.

This post walks through what the shift actually means for a business with, say, 10 to 200 people, an engineering team of two or three or a managed IT partner, and a monthly cloud spend somewhere between ₹20,000 and ₹5,00,000. We cover latency and user experience, data residency and the DPDP Act obligations you're now accountable for, pricing realities in rupees including GST and currency exposure, and a checklist you can run before you move a single workload. There's also a worked example of a Gurgaon logistics company that made the switch, including the parts that didn't go to plan.

Key Takeaways

• Regional capacity in India changes latency and compliance, but it doesn't automatically lower your bill. Egress, NAT gateway charges and cross-region replication still decide your final number.
• Map every workload to its data classification before you move anything. Personal data of Indian users now carries DPDP Act obligations regardless of where the server sits.
• Price in rupees with GST and forex spread included. A dollar-denominated estimate can be off by 4 to 8 percent once you add conversion and tax treatment.
• Benchmark real round-trip latency from your actual user locations (Mumbai, Bengaluru, Delhi NCR, Hyderabad, Tier-2 cities) rather than trusting region marketing pages.
• Move in waves, not in a single cutover. Keep a rollback path for at least 30 days after each wave.
• Get a written migration and egress cost estimate from your provider or partner before committing to a reserved instance or savings plan.

Why AWS and Azure Capacity in India Matters Right Now for SMB Cloud Costs

When hyperscalers add or expand regions, the immediate effect is technical: lower round-trip time to users, more availability zones to design across, and in some cases new managed services that become available locally first. The second-order effect is commercial. Once a region has enough capacity, providers can offer better pricing on committed use, and customers who were previously forced to route traffic through Singapore or Mumbai hubs for certain services can often collapse that path.

The West Asia rerouting story adds a layer of resilience thinking. Organisations that had production workloads hosted in Gulf regions have had to think hard about failover, and many are now evaluating India as a primary or secondary location. For an Indian SMB, this is not a geopolitical concern to ignore. It's a signal that regional diversification is now a board-level question even at small scale.

A practical caution: capacity announcements tell you what is possible, not what is cheaper for your specific shape of traffic. Always verify service availability in the specific region you plan to use, because not every managed service, GPU instance type, or database tier is available in every Indian region on day one.

Does Moving to an Indian Cloud Region Actually Reduce Latency for Indian Users?

For users in Indian cities, yes, usually by a meaningful margin. A request from a user in Bengaluru hitting a server in Mumbai typically sees round-trip times in the low tens of milliseconds, while the same request routed to Singapore adds a substantial penalty that users feel in interactive applications: checkout flows, dashboard refreshes, voice calls, and anything with multiple sequential API calls.

The real impact depends on your architecture. A static website barely notices. A point-of-sale system that makes eight sequential API calls per transaction notices a lot. Chatty applications that were designed on the assumption of a fast regional link tend to show the biggest gains after migration, and also the biggest surprises, because the latency you were hiding is suddenly visible.

How to Measure Latency Before You Decide

1. Pick three to five representative user locations. For most Indian SMBs this means your head office city, your largest customer cluster, and one or two Tier-2 cities where field staff work.
2. From a machine or VM in each location, run ping and mtr to the public endpoint of your current region and to a candidate Indian region's endpoint. Record median and 95th percentile values over at least an hour, during business hours.
3. For your application, measure time-to-first-byte and total page or API completion time from the same locations, not just network RTT. Application round-trips multiply network latency.
4. Repeat measurements across at least two ISPs if you can. Jio, Airtel and a fibre provider often route very differently.

Pro Tip: Don't benchmark from your office LAN alone. Most SMB staff now work from home or from client sites, and the mobile carrier path to the cloud is what your customers and field teams actually experience. Run a test from a 4G/5G phone connection on the same floor as your server room, if you have one, and you'll often see the real bottleneck is the last mile, not the region.

What Do Data Residency and DPDP Compliance Mean for Your Cloud Choice?

Data residency is the question of where data is stored and processed. Data protection is the question of how it is handled. These overlap but are not the same, and Indian SMBs often conflate them.

The Digital Personal Data Protection Act, 2023 and its rules framework bring obligations around consent, purpose limitation, security safeguards, breach notification and rights of data principals. The Act does not, as a general matter, require all personal data to be stored in India. It does allow the central government to restrict transfers to specified countries, and sectoral regulators can impose stricter localisation requirements. So "we moved to Mumbai" is not automatically compliance, and staying in Singapore is not automatically a violation.

Sectoral rules matter more than the general law for many businesses. Payment data is governed by RBI guidelines and PCI DSS obligations. Insurance, telecom, health and certain financial services carry their own localisation and audit expectations. If you're a fintech, an NBFC or a healthtech SMB, check your regulator's current directions before you assume anything.

There is also a practical point. Hosting in India gives you one less thing to explain to a bank's vendor-risk team or an enterprise client's procurement desk. Many large Indian buyers now ask for India-hosted data as a default requirement in vendor questionnaires, even where law doesn't mandate it. That commercial pressure is often the real driver.

A Simple Data Classification Exercise

Before you pick a region, sort your data into four buckets:

• Regulated personal data (KYC, health, payment, employee records with Aadhaar or PAN): needs a documented residency decision and an audit trail.
• Ordinary personal data (customer names, emails, order histories): DPDP obligations apply; residency is a risk decision, not a legal one in most cases.
• Business confidential data (pricing, contracts, source code): residency preferences vary; access control matters more.
• Public or derived data (marketing assets, anonymised analytics): usually low residency concern.

Document the decision for each bucket. When an auditor or a client asks why a particular dataset lives where it does, a one-page register with the reasoning is worth more than any region certificate.

How Do Cloud Costs in India Compare Once You Count Rupees, GST and Egress?

Cloud pricing is published in US dollars for most services, and Indian businesses pay in rupees through a reseller, a direct invoice in INR, or a corporate card with forex markup. The sticker price in a region is only the start. The line items that surprise SMBs most are data transfer out to the internet, cross-availability-zone traffic, NAT gateway processing charges, and inter-region replication fees.

On GST: for most B2B cloud services bought by a registered Indian business, the 18 percent GST is typically charged on the invoice and can usually be claimed as input tax credit, provided the invoice is in your name with your GSTIN and the service is used in the course of business. Verify with your chartered accountant, because treatment depends on whether the supply comes from a foreign entity under reverse charge or from an Indian reseller, and that choice affects both cash flow and reconciliation.

On currency: if your provider bills in USD and your card converts at a spread, your effective cost can drift 2 to 4 percent above the published rate, plus any bank forex markup. Ask whether you can be invoiced in INR directly. Some resellers and partners offer this, sometimes with a fixed rate for the contract term.

Worked Example: A Gurgaon Logistics Company

Consider a 15-person logistics operator in Gurgaon. They ran a dispatch and tracking application on three on-premise servers in a rented rack, plus a backup appliance and a leased line. Their monthly outlay came to roughly ₹45,000, covering power, colocation, leased line rental, hardware amortisation and a part-time sysadmin's time allocated to the system.

Their first cloud estimate, done from the AWS calculator in Singapore, looked attractive at around ₹38,000 per month. But the Singapore region added latency that field dispatchers complained about in the mobile app, and the GPS ping traffic from 60 vehicles generated egress charges their first estimate had missed.

Moving to an Indian region, they re-architected three things. They replaced the chatty per-location API with a batched endpoint, which cut round-trips by roughly 70 percent. They moved the static dashboard assets to a CDN with an Indian edge. And they right-sized the database from a larger instance to a smaller one with read replicas, after two weeks of production metrics showed the original sizing was driven by a one-off data import.

The resulting monthly run-rate settled near ₹18,000 including backups, monitoring and a managed support retainer, with the leased line decommissioned. The migration itself cost roughly ₹2.2 lakh in partner time over eight weeks, which took about nine months to pay back. The lesson they reported was not "cloud is cheap." It was that the savings came from re-architecting and from dropping the colocation contract, not from the region switch alone.

Common Mistake: Teams often compare on-prem cost to a cloud estimate that assumes the same architecture running 24/7. In practice, a large share of SMB workloads (reporting jobs, staging environments, dev databases) can be scheduled off outside business hours, and that single change often matters more than the choice between AWS and Azure.

Which Regional Cloud Option Fits Your Business? A Comparison

The right choice depends on your workload, your team's skills, and your existing Microsoft or Google commitments. The table below compares the common paths an Indian SMB evaluates. Treat it as a starting frame, not a verdict, and check current service availability in the specific Indian region.

Option	Typical fit	Strengths for Indian SMBs	Watch-outs
AWS (Mumbai / Hyderabad regions)	Broad workloads, startups, custom-built apps	Wide service catalogue, mature partner ecosystem, many Indian SI partners	Egress and NAT charges can surprise; service parity per region varies
Azure (Central India / South India)	Microsoft 365 and Windows-heavy shops, .NET workloads	Strong fit with existing Microsoft licensing; hybrid tooling	Licensing rules and Hybrid Benefit terms need careful reading
GCP (Mumbai / Delhi regions)	Data and analytics-heavy teams, Kubernetes-first shops	Strong data tooling; competitive pricing on some compute	Smaller partner bench in some Indian cities
Indian cloud or sovereign-leaning providers	Regulated sectors wanting domestic operators	Domestic support, local invoicing in INR, simpler contracts	Narrower service catalogue; check exit and portability terms
Stay on-prem or colocation, hybrid for DR only	Stable, predictable workloads with a strong sysadmin	Full control; no egress surprises	Hardware refresh, power, and DR effort fall on you

If you're deciding between a sovereign or Indian-operated option and a hyperscaler, the trade-offs are explored more fully in our piece on whether data localisation is worth the cost, and the wider provider comparison is in our cloud provider selection guide.

What Hidden Costs Should You Check Before Switching Cloud Providers?

Switching providers or regions has costs that don't appear on any pricing page. Budget for them explicitly, because they are usually the difference between a migration that pays back in a year and one that drags on for three.

• Data transfer out during migration. Moving terabytes out of your old provider can incur egress fees at their rate. Ask both providers for a transfer estimate, and consider whether a physical transfer appliance makes sense.
• Dual running. You'll pay for old and new environments in parallel for weeks. Plan for at least one to two months of overlap for anything customer-facing.
• Reserved and committed-use lock-in. Savings plans and reserved instances reward commitment but penalise early exit. Don't buy a three-year commitment before the architecture stabilises.
• Re-certification and audit. Client security questionnaires, ISO 27001 scope, and sectoral audits may need updating for a new region.
• Partner and tooling changes. Monitoring, backup and CI/CD integrations often need rework. Factor in engineering hours, not just infrastructure.
• GST and invoicing reconfiguration. New invoices, new billing accounts and new vendor master records in your ERP and accounting software.

How Should You Run a Cloud Region Evaluation Step by Step?

Here is a practical sequence you can brief to your IT partner or run in-house. It assumes you already have some workloads in production and are considering a move to an Indian region, either within your current provider or to a different one.

1. Inventory workloads. List every application, database, storage bucket, scheduled job and third-party integration. Note owner, traffic pattern, and business criticality.
2. Classify data. Apply the four-bucket exercise described above and record the residency and protection decision for each.
3. Measure current costs by line item. Pull 90 days of billing data. Separate compute, storage, egress, NAT, managed services and support fees.
4. Measure latency from real user locations. Use the method outlined earlier, including mobile carrier paths.
5. Verify service availability. Confirm every required managed service and instance type exists in the target Indian region. Note any gaps and workarounds.
6. Build a three-scenario cost model in INR. Model current-architecture lift-and-shift, re-architected with right-sizing and scheduling, and a hybrid option. Include GST treatment, forex spread, egress and migration one-off costs.
7. Run a pilot. Migrate one non-critical workload end to end. Measure actual latency, cost and operational effort against the model.
8. Plan waves. Group workloads by dependency. Move low-risk, stateless services first and databases last, with a documented rollback for each wave.
9. Cut over and monitor for 30 days. Keep the old environment warm during this window. Review costs weekly.
10. Decommission and document. Remove old resources, cancel contracts, update your data register and compliance documentation.

Pro Tip: Ask your provider for a "cost anomaly" alert on day one of the new region. Most cost surprises in the first three months come from one forgotten dev environment, a misconfigured NAT gateway, or a backup job running in the wrong region. Anomaly alerts catch these in days instead of at invoice time.

What Should an Indian SMB Ask Its Cloud Provider or Partner Before Signing?

Whether you buy direct or through a partner, the questions below will surface the answers that matter. Ask for written responses.

• Which of my current services are available in the target Indian region today, and what is the roadmap for the rest?
• What is the total egress estimate for my migration, and what is the egress rate for ongoing traffic?
• Can I be invoiced in INR, and how is GST applied to my invoice? Is it reverse charge or domestic?
• What are the data processing and sub-processing locations, and how are support staff access controls handled?
• What is the exit process, including data export format and timelines?
• What support SLAs apply in the Indian region, and what escalation path exists for a production incident at 2 a.m.?

Frequently Asked Questions

Is AWS Mumbai region cheaper than Singapore for Indian businesses?

Compute and storage prices vary by region and change over time, so check the current rate card for your specific instance types. The bigger cost driver is usually egress and the architectural effects of latency, which often favour an Indian region for Indian users. Build a model with your actual traffic before deciding.

Do I need to move my data to India to comply with the DPDP Act?

The DPDP Act does not, as a general rule, require all personal data to be stored in India. It does impose obligations on how data is handled and allows government restrictions on certain transfers. Sectoral regulators such as RBI may impose stricter requirements, so check the rules that apply to your industry.

How much does it cost to migrate a small business from on-premise to cloud in India?

For a small business with a few servers and a handful of applications, migration partner effort often ranges from a few lakh rupees to somewhat more, depending on complexity and re-architecture. Dual running and egress add to this. A written estimate from your partner, broken down by phase, is the right way to get a real number.

Can I pay AWS or Azure in Indian rupees with GST invoices?

Options vary by provider and by whether you buy direct or through a reseller or partner. Many Indian resellers invoice in INR with GST. Confirm invoice currency, GST treatment and whether input tax credit can be claimed with your chartered accountant before you sign.

Will switching to an Indian region break my existing integrations?

It can, mostly through hard-coded endpoints, region-specific ARNs or resource IDs, and managed services that are not yet available in the target region. A dependency inventory and a pilot migration will reveal most of these before cutover.

How long does a cloud region migration take for a 20-person company?

A typical timeline for a modest estate is six to twelve weeks, including inventory, pilot, waves and a 30-day stabilisation period. Simple static or stateless workloads can move faster; anything with a large database or strict downtime windows takes longer.

Should I wait for more regions before deciding?

Waiting has a cost too, particularly if your users are already experiencing latency. A reasonable approach is to decide on workloads that are clearly better in India now, keep flexibility on the rest, and revisit quarterly as service availability and pricing evolve.

Where eDarpan Fits In

If you're working through this evaluation and want a second pair of experienced eyes, eDarpan's team helps Indian SMBs with cloud migration and managed services, and with broader IT consulting when the right answer is not yet clear. We also handle licensing questions around Microsoft 365 and Google Workspace, which often shape the cloud decision more than the infrastructure itself.

Cloud regions are one piece of a broader stack. If you're rebuilding customer-facing systems at the same time, our teams build custom software and mobile apps, and we can connect your new environment to tools like WhatsApp Business API or bulk SMS services without re-engineering everything. If you want to see how a migration plan would look for your business, get in touch or read more about how we work.

The short version: check latency from your real users, classify your data before you move it, model costs in rupees including egress and GST, and move in waves with a rollback plan. A region change can save money and improve experience, but only if you do the unglamorous work of measuring first.

Image credit: car-finance-after-GFC by natloans via flickr (BY-ND 2.0), sourced through Openverse.