Cloud Migration as a Service

Know what you rent and what you must still protect. You will separate the duties a provider holds from those you keep, so no security task is lost in the handover to rented services.

• Models: infrastructure (rented machines), platform (managed tools) and software (ready-made apps), each sold as a service.
• The provider secures the building and hardware; you secure accounts, data and settings.
• Public, private and mixed (hybrid) clouds.
• When not to move: a need for very fast response, special rules, tiny savings.
• Draw a two-column chart of what the provider looks after, such as buildings and hardware, and what you look after, such as accounts, settings and data.
• Compare three service types: rented machines you manage, a managed platform you configure, and ready software you simply use, noting which duties shift to you in each.
• Write down which cloud services you already use without knowing, such as free email or file sharing.
• Note the exit route: how you would leave a provider if prices rise.

Activity. Draw a responsibility table: provider versus you for five items.

Standard to meet. Check that the table is correct for each model.

Common mistake. Assuming the provider secures everything.

Check yourself. List three things you still secure after the move.

You cannot move what you have not listed. You will build an application and dependency inventory that shows which systems must move together and which can safely wait.

• List servers, apps, data size, users, licences and links to other systems.
• Mark what must run together.
• Find hidden items: scheduled jobs, printers, old shares.
• Record owner and business importance.
• List every application, its owner, its data size in gigabytes, its users and the other systems it calls; aim for one row per item in a simple sheet.
• Trace dependencies by asking what breaks if each item stops, and group items that must move together so you do not split a pair across two places.
• Interview users about which systems they cannot work without, and rank them.
• Note licence terms; some software cannot move without a new licence.

Activity. Build an inventory of at least 8 items with owners.

Standard to meet. Draw the dependencies and check them with the owner.

Common mistake. Forgetting hidden links such as a nightly job.

Check yourself. Which hidden items did your inventory find?

Lift as-is, tweak, or rebuild: each has cost and risk. You will assign each workload to a move, adjust, rebuild, replace or retire path and choose a low-risk pilot to learn from first.

• Rehost (as-is) is quick; replatform (small change) saves more; rebuild costs most.
• Compare on cost, time, risk and skill needed.
• Pick by business need, not by fashion.
• Decide a move order: low risk first.
• Sort each workload into move as is, adjust slightly, rebuild, replace with ready software, or retire, and give a one-line reason for each choice.
• Start with a low-risk, low-value workload as a pilot so the team learns the process before touching anything the business depends on.
• Run a pilot with a low-risk system first, such as shared files.
• Record skills needed and who will learn them.

Activity. Score two options on a one-to-five table.

Standard to meet. Justify the choice with scores and a reason.

Common mistake. Choosing a rebuild when a rehost would do.

Check yourself. Why did you choose your option?

Cloud bills surprise people who skip the sums. You will estimate a monthly cloud bill from hours, storage and transfer, add a buffer, and set alerts so spending never becomes a surprise.

• Count compute, storage, data out, support and backups.
• Use the provider's free calculator, then add 20% margin.
• Compare with current costs over 12 months, including staff time.
• Set budget alerts.
• Estimate monthly cost as machine hours, storage in gigabytes, data transfer out and support, and compare with your current yearly cost divided by twelve.
• Add a 20 percent buffer for items you forgot, and set a budget alert at 80 percent of your monthly limit so overspend is noticed early.
• Include staff time for the move and for training.
• Compare reserved and pay-as-you-go prices for steady loads.

Activity. Build a 12-month cost table.

Standard to meet. Show all cost lines with currency and include margin.

Common mistake. Comparing prices without data-out and support.

Check yourself. Which cost line surprised you?

Move the data safely and keep it private. You will classify data, decide where each class may live, and secure admin accounts, so the move does not weaken protection.

• Strong sign-in with a second step; least access.
• Encrypt data (scramble it) while it moves and while it is stored.
• Know where data is stored; follow local data law.
• Log who did what.
• Classify data into public, internal and confidential, and decide for each class where it may be stored and who may reach it.
• Turn on multi-step sign-in for every admin account on day one and use separate accounts for daily work and for administration.
• Check the provider's location for data and the law that applies.
• Set automatic alerts for odd sign-ins and large downloads.

Activity. Write a security checklist and tick it for your workload.

Standard to meet. Complete the checklist and fix one gap.

Common mistake. Skipping the second sign-in step for admins.

Check yourself. How many accounts lack a second step now?

A test move shows problems when they are cheap. You will restore a backup and rehearse the whole move on a copy, so timings and failures are known before real systems are touched.

• Copy a small slice first.
• Restore a backup once to prove it works.
• Time the move: size divided by speed.
• Write the cut-over steps minute by minute.
• Restore a backup into a clean area and open the files; a backup that has never been restored is only a hope.
• Rehearse the full move on a copy at least once and time each step, so you know whether the cut-over fits inside the agreed window.
• Rehearse the rollback as well as the move.
• Keep a list of who must be called if a test fails.

Activity. Run a trial move and a restore test.

Standard to meet. Restore works; time measured.

Common mistake. Never restoring a backup.

Check yourself. How long did your restore take?

Move in a quiet window with a way back. You will write a timed cut-over script with numeric rollback triggers, so a failed move can be reversed quickly and without argument.

• Tell users in advance.
• Freeze changes, move, test, switch.
• Decide a go or no-go point; roll back if tests fail.
• Keep the old system for an agreed period.
• Write a numbered cut-over script with times, owners and checks, and agree a go or no-go point, such as 30 minutes before the window closes.
• Define rollback triggers in numbers, for example error rate above 2 percent or login failures over 10, and keep the old system untouched until day seven.
• Send two user notices: one week and one day before.
• Keep a short list of known issues ready for the help desk.

Activity. Write cut-over and rollback plans.

Standard to meet. Give plans owners, times and go or no-go rules.

Common mistake. No rollback rule.

Check yourself. What is your no-go rule?

Prove the move worked and teach the owner. You will compare results with a baseline after thirty days and give the owner a runbook, so the benefits are proven and the system is theirs.

• Compare speed, cost, complaints and downtime with goals.
• Update the inventory and costs.
• Train users and write a one-page guide.
• Plan 90 days of improvements.
• Compare after 30 days against the baseline for speed, uptime, cost and user complaints, and present the four numbers on one page.
• Hand over a runbook with accounts, backups, alerts and contacts, and run a short walk-through so the owner can do each task without you.
• Review the first bill against the plan within 30 days.
• Schedule a six-month check for unused resources and security settings.

Activity. Write a results page and a 90-day plan.

Standard to meet. Compare results to targets using numbers.

Common mistake. Declaring success without numbers.

Check yourself. What are your three measured results?

A firm plans a move on Friday afternoon with only one technician.

Question: What is the risk?

Model answer: Problems appear when staff are leaving and help is thin. Move in a quiet window with two people and a rollback plan.

A trial bill is three times the estimate.

Question: What do you check?

Model answer: Data egress, always-on test machines and storage tiers; add budget alerts and shut down unused resources.

On the slide: L1 Basic · Basics of cloud

Notes: Explain cloud storage, computing and software as a service in your own words and name one thing each is good for. Notes: move one folder to a free file service, share it with a friend and write what you can and cannot control. Standard: three sentences that a non-technical friend can repeat.

On the slide: L2 Awareness · Awareness

Notes: Describe the three service models and list five risks of moving a workload: lock-in, outage, cost surprise, data location and weak settings. Notes: compare three real services from their public pages without naming them, in a table of price, location of data and exit terms. Standard: every cell has a source and a date.

On the slide: L3 Foundation · Foundation

Notes: Inventory one small setup (a website, shared files or a small app): what it holds, who uses it, how big it is and what it costs now. Notes: spreadsheet only, no live system touched. Standard: a list a colleague could use to size the move, with units and currency.

On the slide: L4 Practitioner · Practitioner

Notes: Run a small trial move of practice data and measure the cut-over time, the cost for the month and one failure you caused on purpose. Notes: use practice data only, back it up first and write the rollback step before you start. Standard: the trial log shows times, cost and a successful rollback.

On the slide: L5 Specialist · Specialist

Notes: Plan and complete the trial migration of one real small workload, with a cost comparison, security settings, a cut-over window and a tested rollback. Notes: the plan names an owner for each step, a go or no-go check and the data location. Standard: cut-over inside the window, rollback proven, monthly cost compared with the old setup.

On the slide: L6 Advanced · Advanced

Notes: Lead a move of several linked systems, with a dependency map, a staged cut-over and a governance record of who approved what. Notes: add automation for repeated steps and a cost alert; review the plan with a second engineer. Standard: no system moves before its dependencies, and the budget alert is tested.

On the slide: L7 Super expert · Super expert

Notes: Write the migration playbook other teams reuse and train the trainers who run it. Notes: hold a review after each move and fold its lessons, such as cut-over times and rollback triggers, into the playbook.