📖 The Beginner's Guide to Enterprise Cloud Migration

Table of contents

🧭 The four phases

What: Every large migration moves through Discovery (what do we have?), Planning (what do we do with each thing?), Execution (move it) and Operations (run it in the cloud). The Cloud Architect coordinates Business Systems Analysts (BSAs — people who translate between business needs and IT), developers, QA (Quality Assurance testers), Ops (operations/infrastructure engineers), network teams and executives.

How:

1. Discovery: inventory servers (CMDB = Configuration Management Database, the list of all IT assets), collect 30–90 days of utilisation, map dependencies.
2. Planning: build the business case, classify by the 6 Rs, size the team, build the landing zone design.
3. Execution: landing zone → pilot → waves of replication and cutover.
4. Operations: hypercare, cost optimisation, security posture, decommission.

Analogy: Moving house. First you inventory what you own (Discovery), decide what to keep/sell/bin (Planning), hire the truck and move on a date (Execution), then unpack and change your address everywhere (Operations).

💰 Business case & TCO

What: TCO (Total Cost of Ownership) adds up every cost of running servers over several years. On-prem is mostly CapEx (Capital Expenditure — buying hardware up front, plus cooling, power, real estate). Cloud is OpEx (Operating Expenditure — pay-as-you-go per hour). The business case compares them and asks the CFO (Chief Financial Officer) for money.

How:

1. Export the CMDB and 90-day CPU/memory telemetry.
2. Flag zombies (<1% CPU) and end-of-life (EOL) systems for retirement.
3. Model 5 years: on-prem refresh spikes vs cloud hourly × (1 − reserved discount).
4. Add dual-running costs, egress, licensing, training.
5. Write assumptions explicitly; present savings % against the company's hurdle rate.

Analogy: Comparing owning a car (buy it, insure it, garage it, replace it in 5 years) with a ride-share subscription — and first noticing you own three cars you never drive.

🗂️ The 6 Rs

What: Six strategies for each application. Rehost (Lift & Shift: copy the VM as-is using AWS Application Migration Service — MGN — or Azure Migrate). Replatform (Lift, Tinker & Shift: small change for big benefit, e.g., Oracle on a VM → Amazon RDS or Azure SQL managed database). Refactor (re-architect, e.g., monolith → Kubernetes microservices). Repurchase (Drop & Shop: replace with SaaS such as Microsoft 365). Retire (switch off). Retain (keep on-prem for now — mainframe, latency, compliance).

How:

1. Plot each app: debt (old OS, unsupported DB, no tests) vs value (revenue, users).
2. Low debt → Rehost. Managed equivalent exists → Replatform. High value + high change rate → Refactor (few!). Commodity → Repurchase. Unused → Retire. Anchored → Retain.
3. Sum cost and effort; check against budget and deadline; iterate.

Analogy: Moving house again: most boxes go in the truck unchanged (Rehost), the old fridge is replaced by the landlord's (Replatform), the kitchen gets remodelled after you move in (Refactor), you subscribe to a streaming service instead of moving the DVD shelf (Repurchase), half the garage goes to the tip (Retire), and the piano stays with your parents for now (Retain).

🚚 Moving petabytes

What: A petabyte (PB) is 1,024 terabytes. Moving 5 PB over a 10 Gbps (gigabits per second) private link — AWS Direct Connect or Azure ExpressRoute — takes 50+ days at realistic utilisation. Alternatively ship physical appliances: AWS Snowmobile (a 100 PB truck) or several Azure Data Box Heavy units (≈1 PB each).

How:

1. Compute days = (PB × 1024 TB) ÷ TB-per-day; compare with the deadline.
2. If too slow, order appliances; book the loading dock; plan the 3-day local copy.
3. Ship with chain of custody; ingest in the cloud; verify checksums; wipe devices.
4. Run incremental sync over the wire until cutover.

Analogy: Emailing 5,000 photos one at a time versus posting a hard drive.

🔀 Cutover weekend

What: Cutover is the moment users switch from old to new. Developers update CI/CD (Continuous Integration/Continuous Delivery) pipelines and change database connection strings from IPs to cloud DNS names. Ops run Terraform IaC (Infrastructure as Code) and the final delta sync. QA run data-integrity, smoke and load tests. Network Ops flip the DNS GTM (Global Traffic Manager) CNAME record so the website name points at the cloud.

How:

1. Fri 18:00 Go/No-Go #1; 20:00 stop traffic.
2. Dev: pipelines + connection strings. Ops: terraform apply; Sat 02:00 final delta.
3. QA: row counts/checksums, smoke, load (p95 ≤ 300 ms).
4. Sat 09:30 Go/No-Go #2 → 10:00 DNS switch (TTL 60 s).
5. Sun 12:00 business sign-off → hypercare.

Analogy: Changing planes mid-flight — you prepare everything, then switch in one coordinated moment with a parachute (rollback) ready.

🛒 Globex case study

What: A fictional e-commerce company with a Java monolith on 40 VMware hosts, a lease ending in 18 months and $2M/hour outage cost. The story follows BSAs finding hidden dependencies, developers containerising, QA baselining and Ops implementing CSPM (Cloud Security Posture Management — tools that continuously check cloud configuration for risks).

How: Scroll the page; each chapter animates; click components for detail.

Analogy: A documentary about one family's house move rather than a moving-company brochure.

🧱 IaaS, PaaS, SaaS & hypervisors

What: A hypervisor (VMware ESXi, Microsoft Hyper-V, AWS Nitro) is software that splits one physical server into many virtual machines (VMs). IaaS (Infrastructure as a Service) rents you VMs — you manage the operating system. PaaS (Platform as a Service) rents you a managed database or runtime — the provider patches OS and engine. SaaS (Software as a Service) rents you the finished application — you manage identity and data policy only.

How:

1. ESXi/Hyper-V → EC2/Azure VM via block-level replication (MGN/Azure Migrate).
2. Oracle RAC → RDS/Azure SQL via DMS (Database Migration Service) + schema conversion.
3. Exchange → Microsoft 365 via hybrid mailbox moves; retire the farm.

Analogy: Owning a house (on-prem), renting an apartment (IaaS), staying in a serviced apartment (PaaS), booking a hotel (SaaS).

👥 The 25-person team

What: A Migration Center of Excellence (MCoE) — a core team that owns patterns, tooling and delivery. 2 Enterprise Cloud Architects (landing zone, IAM — Identity and Access Management — guardrails, Transit Gateway routing), 3 PMs/Scrum Masters (T-minus plan, stand-ups, CAB — Change Advisory Board), 8 Migration Engineers (agents, sync, Terraform/Bicep), 4 BSAs (dependencies, owners, outage windows), 4 QA/Automation (integrity, load, connectivity), 4 Day-2 SREs (Site Reliability Engineers — post-cutover operations).

How: Size engineers from servers ÷ months ÷ ~35 servers per engineer-month; keep QA at 1 per 2–3 engineers; never go below 1 architect or 2 SREs.

Analogy: A film crew — a few directors, many camera operators, and someone must own post-production or the film never ships.

🏛️ SteerCo & governance

What: The Executive Steering Committee (SteerCo) of CIO (Chief Information Officer), CFO and CISO (Chief Information Security Officer) makes go/no-go decisions, accepts risk and releases budget. The CISO needs on-prem firewall rules (Palo Alto, Cisco) translated into cloud NSGs (Network Security Groups) and a WAF (Web Application Firewall). The CIO needs a burn-down chart (servers remaining vs time). The CFO needs spend vs progress.

How: Present security parity → schedule → spend; record decisions and actions; escalate through workstream lead → PMO → SteerCo → Board.

Analogy: A planning committee for a building: the safety inspector, the schedule manager and the treasurer all have a veto.

⏱️ T-minus timeline

What: A countdown plan from T-12 months (Discovery with Movere/Cloudamize), T-6 (landing zone + 50-server pilot), T-4 weeks (replication starts), T-0 (48-hour cutover: Fri 8 PM stop traffic, Sat 2 AM delta, Sat 10 AM DNS, Sun 12 PM sign-off), T+1 to T+14 days (hypercare war room).

How: Fix T-0 per wave; lay out predecessor tasks; add a dress rehearsal at T-1 week; keep a 2-week hypercare buffer.

Analogy: A rocket launch countdown — every team has a call-out at a fixed T-minus time.

📊 Comparison tables

The 6 Rs compared

Wire vs truck for 5 PB

Shared responsibility

🧪 Interactive explorer

Click a concept chip. The flow diagram and step list rebuild for that concept and animate.

🎯 Decision guide

🔤 Glossary