Career OS

Case Study — Reading the ayrisglobal Stack

Six lectures of theory, now applied to one real production setup: yours. By the end you’ll be able to read any company’s infrastructure off a few terminal commands — and you’ll understand the email-migration decision from the ground up.

⏱ ~15 min · 🔴 Advanced · Prerequisites: Lectures 1–6

🎯 What you’ll be able to do

  • Map a real domain’s registration, DNS, hosting, and email from raw DNS records
  • Explain why each piece is where it is
  • Reason through an infrastructure migration without breaking things

Step 1 — What we can observe

A few commands reveal the whole stack (you ran versions of these already):

nslookup -type=ns  ayrisglobal.com          # who runs DNS
nslookup -type=mx  ayrisglobal.com           # where email goes
nslookup -type=mx  assure.ayrisglobal.com    # subdomain email
curl -I https://ayrisglobal.com              # who fronts the website

What they told us:

LayerFindingLecture
Registrarayrisglobal.com registered at Hostinger (expires 2027-05-31)2
DNS / authoritative NSCloudflare (lucy/elliott.ns.cloudflare.com)3
Website frontCloudflare (CDN/proxy) → origin host4, 6
Email (MX)Hostinger (mx1/mx2.hostinger.com) for root and assure.3, 5

Step 2 — The stack, drawn out

                         ┌─────────────────────────────────────────┐
   Registrar: Hostinger  │  owns the NAME ayrisglobal.com            │
   (just the lease)      └───────────────────┬─────────────────────┘
                                              │ NS records delegate DNS to ↓
                         ┌────────────────────▼─────────────────────┐
   DNS + CDN: Cloudflare │  source of truth for all records          │
                         │  A/CNAME → website   ·   MX → email        │
                         └───────┬───────────────────────┬───────────┘
                                 │ web traffic            │ mail (MX)
                                 ▼                        ▼
                     Cloudflare edge → origin      Hostinger mail servers
                     (static frontend, fast)       (7 mailboxes @assure.)

The key insight this proves

Registration (Hostinger), DNS (Cloudflare), website host (Cloudflare), and email (Hostinger) are four independent choices. That’s not a mess — it’s normal, and it’s why you can change email without touching anything else. The only thing you’ll edit is MX + a few TXT records inside Cloudflare (the authoritative DNS from lecture 3).

Before you read on, prove you can actually read this stack — not just recognize the words:

Step 3 — The actual situation (from the dashboards)

  • One registered domain: ayrisglobal.com (auto-renew currently OFF ⚠️ — lecture 2 says fix this).
  • All mail lives on the subdomain assure.ayrisglobal.com, on Hostinger’s “Free Business Email” plan expiring 2026-06-29 ⚠️.
  • 7 mailboxes, 1 GB each: darshan, kuldeep, jay, support, krishna, shiparajsolutions, sales — all @assure.ayrisglobal.com.

Step 4 — The migration decision, reasoned from first principles

Goal: move email off Hostinger, keep the addresses, stay cheap and low-maintenance.

Constraint (lecture 3): one MX set per domain → one inbound provider. So we pick one provider that does mailboxes and aliases and forwarding for assure.ayrisglobal.com.

Mailbox vs alias (lecture 5): of the 7 addresses, only the real people need paid mailboxes; role addresses (sales@, support@, …) can be free aliases. → confirm who’s who before paying for 7.

The plan:

NeedChoiceWhy
Human mailboxes + aliasesZoho Mail Lite (~$1/user/mo)Free tier blocks IMAP + multi-domain; Lite is still far cheaper than Google ($6)
App / transactional mailResend (free tier) → AWS SES at scaleNever send app mail through the mailbox provider (lecture 5)
IsolationApp mail from a subdomain (send.)Protects human-mail reputation (SPF/DKIM/DMARC, lecture 5)

The safe cutover (lectures 2, 3, 5):

1. Lower MX TTL in Cloudflare (a day ahead)          ← lecture 3
2. Create mailboxes + aliases in Zoho; verify domain (TXT)
3. Migrate old mail (Zoho's IMAP import) — DON'T delete Hostinger yet
4. Switch MX → Zoho in Cloudflare; add SPF + DKIM + DMARC TXT records
5. Test delivery for a few days
6. Only then cancel Hostinger email
7. (Separately) wire Resend for app mail on a send-subdomain
8. While you're here: turn ON auto-renewal for ayrisglobal.com  ← lecture 2

The one rule that prevents disaster

Never delete the old mailboxes until the new MX is verified and mail is flowing. Email in flight during a botched cutover is simply lost. Overlap, verify, then decommission.

✅ Check yourself

From the records, where is the domain registered vs where is its DNS run — and why isn't that contradictory?

Registered at Hostinger, DNS run by Cloudflare (NS records delegate it there). Not contradictory: registration and DNS are independent layers (lecture 2). The registrar just holds the lease; whoever the NS points to is authoritative for records.

Why can't we keep some mailboxes on Hostinger and move others to Zoho on the same subdomain?

Because assure.ayrisglobal.com has one MX set → one inbound provider for all its mail (lecture 3). Splitting would require different domains/subdomains.

Why send transactional mail from `send.ayrisglobal.com` instead of the main domain?

To isolate its sending reputation and DNS auth records, so high-volume app mail can’t damage the deliverability of the humans’ everyday email (lecture 5).

What's the single most dangerous step in the migration, and how do you de-risk it?

Cutting over MX / decommissioning the old provider. De-risk by lowering TTL ahead of time, running both in overlap, verifying delivery for several days, and only deleting the old mailboxes once the new flow is confirmed.

You run only one command — curl -I https://ayrisglobal.com. Which of the four layers can you NOT learn from it, and what command would you run instead?

curl -I only reveals the website front (Cloudflare). It tells you nothing about the registrar, the authoritative DNS, or email. For DNS run nslookup -type=ns, for email run nslookup -type=mx, and for the registrar you check WHOIS or the registrar dashboard.

Lowering the MX TTL in Cloudflare a day before cutover — what problem does that specifically solve?

TTL controls how long resolvers cache the old MX record. If TTL is high, some senders keep mailing the old provider long after you switch. Lowering it ahead of time means resolvers pick up the new MX almost immediately at cutover, shrinking the window where mail could go to the wrong place.

Of the 7 addresses, which deserve paid mailboxes and which can be free aliases, and why does that distinction save money?

Real people (darshan, kuldeep, jay, krishna) need paid mailboxes because mail is stored and read separately. Role addresses (sales, support, shiparajsolutions) can be free aliases that forward into a real mailbox. You only pay per mailbox, so converting roles to aliases means paying for 4 instead of 7.

📚 Resources

One-line summary

A real stack is just four independent layers — registrar, DNS, web host, email — that you read off a few dig/curl commands. Knowing they’re independent is exactly what lets you migrate email by editing only MX + TXT records in Cloudflare, safely and cheaply.

Still Unclear?

Paste one of these into Claude to go deeper on your own stack:

I ran these and got this output [paste your nslookup -type=ns, -type=mx,
and curl -I results]. Read my stack back to me layer by layer: registrar,
authoritative DNS, website front, and email. Tell me what each command did
and did NOT prove.
Walk me through a zero-downtime email migration from provider A to provider B
for a domain whose DNS is on Cloudflare. List the exact record changes, in
order, with TTL handling, and tell me the one step where mail could be lost.
Explain the difference between a paid mailbox and a free alias for a small
team, and help me decide which of my addresses [list them] should be which
to minimise cost without losing any mail.

Why AI Can’t Do This For You

AI can recite what MX and NS records are, but it can’t run the commands against your domain, read your dashboards, or know that auto-renew is off and the email plan expires in two weeks. Reading a real stack is detective work on live evidence — and the migration decision is a judgement call that balances cost, risk, and who actually needs a mailbox. Get the reasoning wrong and you lose real mail in flight; AI won’t feel that consequence, you will. The skill is connecting raw record output to a safe plan, and that only forms by doing it on a stack you own.

Module done? Add it to today’s tracker

Saves your progress on this device.