Docs · Comparison

CloudNx vs AWS Mumbai

A line-by-line look at how CloudNx maps to every service on a typical Indian-business AWS Mumbai bill. We tell you what's live today, what's coming this year, and the handful of things we deliberately don't sell — so you can decide whether to move with confidence instead of finding out three months in.

This page is built from a real, anonymised AWS India invoice — 15 billed services, USD 786 total. Numbers in the "AWS $" column are the actual line items so you can see where the money goes.

The 15-service audit

Coverage at a glance

Of the 15 services on this bill:

• Live today— EC2, Data Transfer, VPC, ELB, S3. That's 86% of the dollar value ($677 of $786).
• Coming this year — RDS, CloudWatch, ECR, CloudFront, KMS, SNS, ECS. The remaining 14% of the bill, on our public Q3/Q4 2026 roadmap.
• Deliberately not sold— Lambda, Glue, CloudWatch Events. If these are load-bearing for your workload, AWS is the right answer for you and we'll tell you that on the call.

What CloudNx has that the AWS bill above doesn't show

Either the customer wasn't using these on AWS, or AWS calls them something else. All live today:

• Elastic IPs — 1:1 NAT, all-port forwarding, swap on failover
• Firewalls — AWS calls them Security Groups; per-instance ingress/egress rules
• Block volumes — AWS EBS equivalent; NVMe-backed, online resize
• Image catalog — Ubuntu / Debian / Rocky / AlmaLinux / Fedora / openSUSE
• SSH key management — central key store, autorotation hooks
• IAM — sub-users + policies + access keys; SigV4-compatible
• Audit logs — AWS CloudTrail equivalent; DPDP-compliant
• Browser SSH console — AWS Session Manager equivalent, in your browser
• UPI + Razorpay autopay — no Indian rails on AWS at all
• GST tax invoices — proper Indian-entity invoice with your GSTIN, monthly

Where we deliberately won't compete

Lambda + CloudWatch Events (serverless functions + scheduled triggers)

We don't sell serverless functions. The honest reason: doing Lambda right requires a fleet of Firecracker microVMs across many hosts, with a custom scheduler and cold-start engineering that takes AWS years and Cloudflare another year on top of that. We'd ship a worse version and charge you for it. If your workload is event-driven autoscale-to-zero (a webhook receiver, a Slack bot, a cron processor), AWS Lambda or Cloudflare Workers is the right choice. We'll happily run the always-on backend services around them on our side.

Glue (ETL + data catalog)

Out of scope. We're an IaaS company, not a data platform. If you need managed ETL, look at AWS Glue, Databricks, or roll your own on plain VMs (which we're happy to host).

Multi-AZ HA, in the strict AWS sense

Our default deployment is single-host in Helsinki. For an enterprise customer who needs HA, we offer a 2-node Proxmox cluster on dedicated OVH Mumbai hardware — the SLA goes from 99.5% to 99.95%, the price goes up to match. See our Instances docs for the availability model.

The price comparison

Hand the prospect's spreadsheet a like-for-like translation and the picture changes quickly. A 3-VM production stack (1 web, 1 DB, 1 bastion + 200 GB EBS + 1 TB bandwidth):

Half the price, with GST input credit you can actually claim, in a single INR-denominated invoice that ties to your Indian-entity accounts. Notice we lose on the t-family— AWS burstable instances have a CPU-credit pool we don't mirror. If your workload uses many t-class VMs as burst capacity, factor that in honestly.

Who should move

• You're an Indian-registered B2B SaaS, 2-15 engineers, paying AWS Mumbai in USD via credit card
• Your CFO has complained about USD-INR conversion, missing GST input credit, or multiple vendor invoices
• Your workload is web app + Postgres + queue (the "EC2 + RDS" shape), not GPU/ML or serverless-heavy
• Your end-user latency budget is > 50 ms (B2B SaaS, internal tools, cross-border APIs)

Who should NOT move yet

• You need sub-50 ms RTT to Indian end-users today — wait for our Mumbai region (Q3 2026) or stay on AWS Mumbai
• Your workload is heavy Lambda / event-driven serverless
• You need GPU instances for ML training/inference — go AWS, E2E Networks, or Lambda Labs
• You have $100k of unspent AWS Activate credits — burn those first, talk to us when they're done
• Your procurement requires SOC 2 Type II or ISO 27001 today — we're pre-audit and will honestly say so

Next steps

• Read the migration runbook — service-by-service mapping with code samples
• Read the pricing page for the full SKU table
• Sign up — we'll comp ₹500 of credit so you can move a single throwaway workload first and feel the experience

Last reviewed 2026-05-22. Prices and roadmap quarter-targets are re-verified monthly — if you spot a stale number, mail [email protected]and we'll fix it.