Choose the Right Cloud Model

Create a hybrid cloud decision board for a manufacturing company.

Introduction

30 Second Summary

A technology plan can look convincing while overlooking how a business actually operates. One stubborn workload can expose the whole plan.

In this project, you will use draw.io to turn Northstar Manufacturing's workload constraints into a visual recommendation for cloud computing. You will test a public cloud idea before replacing it with a defensible hybrid cloud design.

What You'll Build

Your one-page decision board will reveal the failed first plan beside a final hybrid design that keeps the factory running.

By the end of this project, you'll have:

A before-and-after cloud decision board that makes the failure of the all-public plan visible.
A responsibility map that lets you explain customer control across IaaS, PaaS, and SaaS.
A defensible hybrid target model that places every workload according to its control boundary. You can share it as a PNG export or reopen the editable .drawio source.
Secret Mission: Build a split monthly analytics design that keeps raw sensor data on premises while sending approved summaries to public-cloud analysis.

Are there any prerequisites?

No. You only need the attached training PDF plus a computer with a browser.

Step 1 sets up the free browser workspace. This project creates local diagram files without provisioning paid cloud resources.

Before We Start

Before any hands-on work begins, lock in the decision board you are creating for Northstar Manufacturing. This gives your cloud recommendation a clear business purpose from the start.

Set Up Your Cloud Decision Workspace

Northstar's recommendation needs one visual workspace where the initial assumption can sit beside the improved design. A local draw.io file gives you that workspace without creating any cloud resources.

You will prepare the canvas before adding any workloads or recommendations. Saving the editable file locally also proves that you can recover your work in a later session.

In this step, get ready to:

Open a blank diagram in the browser-based editor.
Create separate areas for the scenario and recommendation.
Save the editable diagram locally and reopen it.

Open a blank local diagram

The editor runs in your browser without requiring an account. Choosing local storage keeps the decision board on your computer.

Enter https://app.diagrams.net in your browser's address bar.
Press Enter to load the editor.
Choose the local storage option when the storage prompt appears.
Click File in the top menu.
Select New.
Select Blank.
Click Create.

You'll see an empty drawing canvas with shape libraries on the left. That blank page is the foundation for Northstar's decision board.

Why Use a Visual Board?

A paper worksheet would also avoid cloud charges. Draw.io gives you an editable file that is easier to revise or share.

Keeping both recommendation versions on one canvas makes each design change visible.

Can't Reach the Blank Canvas?

Confirm that your browser loaded https://app.diagrams.net. Choose the local storage option again if the storage prompt remains open.

If the template window is missing, select File followed by New to reopen it.

Ask for help with the blank diagram setup.

Lay out the decision board

A clear layout separates Northstar's business situation from your recommendation. This structure leaves room for the comparison you build in later steps.

Double-click an empty area near the top center of the canvas.
Select Text from the quick selection box.
Enter Northstar Cloud Decision Board as the title.
Press Enter to save the title text.

You'll see Northstar Cloud Decision Board near the top of the canvas.

Drag a Rectangle from the General shape library onto the left half of the canvas.
Resize the left rectangle using its corner handles.
Label the selected rectangle Scenario by typing the text.
Press Enter to save the label.
Drag another Rectangle from the General shape library onto the right half of the canvas.
Resize the right rectangle using its corner handles.
Label the selected rectangle Recommendation by typing the text.
Press Enter to save the label.

You should now see the board title above separate Scenario and Recommendation areas.

Leave both areas empty for the workload cards you add later.

Labels in the Wrong Place?

Drag each rectangle by its border until Scenario sits on the left. Place Recommendation on the right.

If typing creates separate text, double-click inside the rectangle before entering its label.

Help me fix my decision board layout.

Save and reopen the board

The native .drawio format remains editable when reopened. Closing the board after saving gives you a practical recovery test.

Click File in the top menu.
Select Save as.
Choose the local storage location.
Enter northstar-cloud-decision.drawio as the file name.
Click OK to save the file.

You should see northstar-cloud-decision.drawio in your browser's downloads or saved files list.

Before you reopen the file, which three labels do you expect to see on the canvas?

Close the browser tab containing the decision board.
Enter https://app.diagrams.net in a new browser tab.
Press Enter to return to the editor.
Click File in the top menu.
Select Open From.
Choose the local file location.
Double-click northstar-cloud-decision.drawio in the file picker.
Choose the current-window or new-window option if the editor asks where to open the file.

You'll see Northstar Cloud Decision Board above separate Scenario and Recommendation areas. Both areas remain empty.

That recovery test is complete. Your editable workspace is saved locally and ready for Northstar's cloud analysis.

Can't Reopen the File?

Confirm that you selected the file ending in .drawio. An exported image does not provide the same editable workspace.

Check your browser's downloads list if you cannot find the folder where the file was saved.

Help me locate and reopen my draw.io file.

Your decision board now has a recoverable structure. Next, you will fill it with Northstar's workloads and make the first cloud recommendation.

Create the First-Pass Cloud Design

Your draw.io workspace is saved. You have already proved that it reopens.

Now Northstar needs a visible first recommendation. Moving every workload unchanged into one public cloud looks simple. This baseline makes the assumption visible before you test its hidden trade-offs.

In this step, get ready to:

Map Northstar's four workloads in the Scenario area.
Show the three traditional IT pain points behind the cloud recommendation.
Connect every workload to one Public Cloud destination under Version 1.

Add Northstar's workload cards

A workload card represents one system Northstar relies on. Separate cards keep each system visible when you assess the recommendation.

Drag a rectangle from the shape library into the Scenario area.
Enter Online Storefront inside the rectangle.

You should see the first workload card inside the left side of the board.

Drag another rectangle into the Scenario area.
Enter Legacy Inventory App inside the new rectangle.

You should now see two separate workload cards.

Drag a third rectangle into the Scenario area.
Enter Employee Email inside the new rectangle.

You should now see three separate workload cards.

Drag a fourth rectangle into the Scenario area.
Enter Factory Controller inside the new rectangle.

You should see four distinct systems in the Scenario area.

Save the open northstar-cloud-decision.drawio file.

Show the current-state pain points

Pain cards capture the business pressures behind Northstar's decision. They connect the current environment to the attraction of cloud computing.

Create a pain card labeled slow hardware purchasing below the workload cards.
Create a pain card labeled unused off-peak capacity beside the first pain card.

You should see two current-state problems beneath the workloads.

Create a third pain card labeled too much server maintenance beside the other pain cards.
Save the open northstar-cloud-decision.drawio file.

The Scenario area should now show four workloads plus three reasons for considering cloud computing.

Build the Version 1 recommendation

The first-pass design gives every workload the same destination. Keeping this recommendation simple makes its core assumption easy to inspect.

Add Version 1 as a visible heading in the Recommendation area.
Place a large box labeled Public Cloud below the heading.

You should see one shared destination for the entire first-pass design.

Draw a connector from Online Storefront to Public Cloud.
Draw a connector from Legacy Inventory App to Public Cloud.

You should see the first two workloads following the same route.

Draw a connector from Employee Email to Public Cloud.
Draw a connector from Factory Controller to Public Cloud.

All four workload cards should now connect unchanged to the same destination.

Add an on-demand tag to the Public Cloud box.
Add an elastic tag to the same box.

You should see two expected cloud benefits attached to the recommendation.

Add a shared resources tag to the Public Cloud box.
Save the open northstar-cloud-decision.drawio file.

Why Version 1 Looks Attractive

The on-demand tag captures access when needed. The elastic tag captures capacity that can adjust as demand changes.

The shared resources tag points to pooled infrastructure. These benefits explain why one public-cloud destination feels tempting at first.

Before you scan the full board, how many workload routes do you expect to end at Public Cloud?

Zoom out until the full canvas fits on screen.

The Scenario area should show four workload cards plus three pain cards.

The Recommendation area should show all four workload routes ending at one Public Cloud box under Version 1.

You have the baseline working: every Northstar workload now points to the same public-cloud destination.

Missing a Card or Connector?

Count the four workload cards before checking their connectors.
Move overlapping connector lines until each route is visible.
Compare each typed label with the exact wording shown above.
Help me find the missing card or connector.

Version 1 is clear enough to challenge. Next, you will run Northstar's constraints through it to expose the one-size-fits-all assumption.

Stress-Test the First-Pass Plan

Your Version 1 design is still open in draw.io. It sends every workload unchanged to one public cloud destination.

The board currently treats every workload as though it shares one operating constraint. It also assumes one level of customer control suits them all.

You will test those assumptions against three Northstar requirements. The original recommendation stays visible so its strengths and weaknesses remain clear.

In this step, get ready to:

Add three scenario constraint cards to the board.
Trace each constraint through the existing Version 1 design.
Mark the failed tests with a written diagnosis.

Add the scenario constraints

A constraint describes a condition that the recommendation must handle. Each card gives you a specific test for the first-pass plan.

Create a constraint card labeled holiday traffic rises sharply beneath the workload cards in the Scenario area.
Create a constraint card labeled the factory controller must keep working when the internet link is unavailable beneath the first constraint card.

You should now see one demand constraint plus one connectivity constraint in the Scenario area.

Create a third constraint card labeled the company does not want to patch or maintain email servers beneath the other constraints.

You should now see three distinct tests beside Northstar's existing workloads.

These Constraints Belong to Northstar

The factory connectivity requirement is a teaching condition for this scenario. The email maintenance preference is another Northstar-specific need.

Other organisations can make different choices after testing their own constraints.

Trace each constraint through Version 1

Before you trace the first constraint, which needs do you think the unchanged public-cloud plan can satisfy?

Draw a connector from holiday traffic rises sharply to the elastic tag in Version 1.
Add the outcome label Copes well: elasticity supports rising demand. beside the connector.

The first trace shows where the simple plan performs well. The existing elastic benefit responds to changing storefront demand.

Draw a connector from the factory connectivity constraint to the path carrying Factory Controller into Public Cloud.
Add a red failure mark labeled X Fails: internet loss interrupts the controller path. beside the connector.

The first failure is now visible. The unchanged plan cannot satisfy Northstar's requirement for local factory operation during an internet outage.

Draw a connector from the email maintenance constraint to the path carrying Employee Email into Public Cloud.
Add a second red failure mark labeled X Fails: email server maintenance remains. beside the connector.

The second failure exposes a different problem. Moving the existing email system unchanged leaves Northstar with the maintenance burden it wanted to remove.

Diagnose the failed assumption

The three traces reveal a pattern. One destination can offer a useful benefit while still failing workloads with different control needs.

Add a text box beneath Version 1 containing One deployment model and one service model do not fit every workload..

You should now see the diagnosis beside the original design. The comparison preserves the evidence that produced your conclusion.

What Does the Diagnosis Capture?

A deployment model describes where a cloud environment operates. A service model describes which layers the provider manages.

Your stress test shows that Northstar must consider both choices for each workload.

Select File from the top menu.
Select Save to update northstar-cloud-decision.drawio.

Before you inspect the full board, which parts of the failed assumption do you expect the red marks to expose?

Zoom out in the browser editor until the entire board is visible.
Confirm that all three constraint cards remain visible.
Confirm that every constraint has a traced outcome in Version 1.
Confirm that the board shows at least two red failure marks.
Confirm that the diagnosis reads One deployment model and one service model do not fit every workload..

You should see three constraint cards with visible traces into Version 1. You should also see at least two red failure marks.

The diagnosis should remain visible beside the original all-public-cloud recommendation.

You have turned the first-pass design into a visible warning that Northstar can act on.

Your first-pass plan has met its limits. Next, you will choose a suitable control boundary for each workload.

Match Workloads to Service Models

The stress test exposed the weakness in Version 1. Northstar still carries too much operational responsibility for some workloads.

The factory controller faces a different problem. It needs local placement control during an internet outage.

The next decision assigns a service model to each workload. This matches customer control to the needs you already uncovered.

In this step, get ready to:

Build a responsibility ladder for IaaS, PaaS, and SaaS.
Classify all four workloads with badges and control rationales.
Distinguish service models from deployment models.

Build the responsibility ladder

A control boundary shows which responsibilities stay with the customer. It also shows which responsibilities move to the provider.

How the control boundary shifts

IaaS gives the customer control over the operating system and deployed applications.

PaaS leaves the customer focused on deployed applications and data.

SaaS provides a ready-to-use application while the customer controls its own data and permitted configuration.

Drag three rectangle shapes from the left panel into open space in the Recommendation area.
Stack the rectangles vertically.

You should now see three empty rows with enough room for a model name and its control boundary.

Enter IaaS gives the customer control over the operating system and deployed applications in the top row.
Enter PaaS leaves the customer focused on deployed applications and data in the middle row.

Your first two rows should show how the customer's focus narrows from operating-system control to application delivery.

Enter SaaS provides a ready-to-use application while the customer controls its own data and permitted configuration in the bottom row.

The completed ladder should end with a ready-to-use application in the bottom row.

Classify each workload

These recommendations belong to Northstar's fictional scenario. A different organization could reach different choices from its own constraints.

Add an IaaS badge beside Legacy Inventory App.
Add this rationale beneath its badge: Northstar keeps operating-system control for the legacy application while the provider manages the underlying infrastructure.

Legacy Inventory App should now carry an IaaS badge plus a rationale about operating-system control.

Add a PaaS badge beside Online Storefront.
Add this rationale beneath its badge: Northstar focuses on storefront applications and data while the provider manages the platform.

Online Storefront should now carry a PaaS badge plus a rationale about focusing on the application.

Add a SaaS badge beside Employee Email.
Add this rationale beneath its badge: Northstar uses a ready-to-use email application without patching or maintaining email servers.

Employee Email should now carry a SaaS badge plus a rationale about avoiding server maintenance.

Add a Retain on premises badge beside Factory Controller.
Add this rationale beneath its badge: Northstar keeps the controller on premises so factory operation can continue when the internet link is unavailable.

Factory Controller should now carry a local-placement badge plus a rationale about operating through an internet outage.

Separate management from placement

Service models answer a management question. Deployment models answer a placement and access question.

Add this note beside the responsibility ladder: Service models describe what is managed. Deployment models describe where and for whom the cloud environment operates.

Your browser may ask whether to replace the existing local file. This is expected because you are updating the same editable source.

Select File in the top menu.
Select Save as.
Return to the same local storage location from earlier.
Enter northstar-cloud-decision.drawio as the filename.
Confirm the save.

Before you review the finished classifications, which workload do you expect to keep the strongest local control?

Adjust the canvas zoom until the responsibility ladder and all four workload cards are visible together.

You should see a three-row ladder covering IaaS, PaaS, and SaaS.

Legacy Inventory App should show IaaS. Online Storefront should show PaaS.

Employee Email should show SaaS. Factory Controller should show Retain on premises.

Each workload should include one sentence explaining its control trade-off. The note should distinguish what is managed from where and for whom the environment operates.

You have turned the failed one-size-fits-all plan into a workload-by-workload control map.

Next, you will use those classifications to build Northstar's hybrid target model.

Build and Defend the Hybrid Model

Your workload-by-workload choices corrected the control mismatch exposed by Version 1. The board now has a defensible service-model choice for every workload.

The failed all-public plan still needs a deployment model that preserves local factory operation. In this step, you will use draw.io to turn those choices into a hybrid cloud design.

In this step, get ready to:

Build the Version 2: Hybrid Cloud target model.
Add an Exam Defense panel containing the five NIST characteristics.
Export the completed board as a PNG image.

Build the hybrid target model

A hybrid cloud binds distinct environments into one operating design. Northstar keeps the controller local while cloud-suited workloads gain public-cloud flexibility.

In the Recommendation area, add a large rectangle beneath Version 1.
Enter Version 2: Hybrid Cloud at the top of the new panel.
Add a box titled On-Premises / Private Environment inside the new panel.
Place a copy of the existing Factory Controller card inside the local environment box.
Keep the Retain on premises badge attached to the copied card.
Add a second box titled Public Cloud inside the new panel.

Why the Boundary Matters

The local boundary gives Factory Controller a local operating home. The public-cloud boundary gives each suitable workload access to cloud services.

The connection between those environments turns two locations into one operating design.

Place a copy of Online Storefront with its PaaS badge inside the public-cloud box.
Place a copy of Legacy Inventory App with its IaaS badge inside the public-cloud box.
Place a copy of Employee Email with its SaaS badge inside the public-cloud box.
Draw an arrow from the local environment box to the public-cloud box.
Label the arrow Inventory Data Sync.

Before you inspect the full panel, which workload should be the only one outside the public-cloud box?

Zoom out until the complete Version 2: Hybrid Cloud panel is visible.

What Should I See?

The Factory Controller card sits inside On-Premises / Private Environment.
The Online Storefront card shows its PaaS badge inside Public Cloud.
The Legacy Inventory App card shows its IaaS badge inside Public Cloud.
The Employee Email card shows its SaaS badge inside Public Cloud.
The Inventory Data Sync arrow connects the two environments.

That is the decision turning point. Your board now protects factory continuity without giving up public-cloud flexibility.

Add the exam defense

A recommendation becomes defensible when each placement connects to a recognized cloud characteristic. The Exam Defense panel turns standards language into visible business reasoning.

Add a new rectangle titled Exam Defense beside the Version 2 panel.
Add On-demand self-service to the panel.
Add Broad network access to the panel.
Add Resource pooling to the panel.
Add Rapid elasticity to the panel.
Add Measured service to the panel.

Your panel now holds the complete five-characteristic set. One business link makes the list useful to Northstar.

Add the business benefit capacity expands for holiday traffic near the Exam Defense panel.
Draw a connector from Rapid elasticity to the new business benefit.

Before you check the connector, which characteristic most directly supports Northstar's short holiday traffic surge?

Trace the new connector from the Exam Defense panel to its business benefit.

You should see Rapid elasticity connected to capacity expands for holiday traffic.

How This Defends the Model

The connector proves that rapid elasticity answers Northstar's holiday demand. The complete set keeps the business recommendation anchored to NIST terminology.

Add the standards note and export

NIST defines Community cloud alongside the three models emphasized here. Huawei Cloud's current fundamentals page focuses on public cloud, private cloud, and hybrid cloud.

The standards note keeps the board accurate across both taxonomies.

Add this note beneath the Exam Defense panel: NIST also defines Community cloud; Huawei's current fundamentals page focuses on Public, Private, and Hybrid cloud..
Read the note on the canvas to confirm every word is visible.

You should see the standards note distinguish NIST's broader taxonomy from the Huawei-focused lesson.

Replacing the existing local copy is safe. The updated save retains all previous board content plus the final hybrid model.

Select File in the top menu.
Select Save as.
Choose the same local storage location you used earlier.
Save the editable source as northstar-cloud-decision.drawio.
Approve the replacement prompt if your browser displays one.

Your editable source now contains the tested Version 1 design plus the defended Version 2 model.

Exporting may open an options window before the image downloads. The PNG copy captures the finished board for viewing outside the editor.

Select File in the top menu.
Select Export as.
Select PNG.
Keep the default export settings in the export window.
Save the exported image locally as northstar-cloud-decision.png.

Why Keep Both Files?

The northstar-cloud-decision.drawio file remains editable. The northstar-cloud-decision.png file provides a shareable snapshot of the finished recommendation.

Before you open the export, do you expect the failed Version 1 design to remain visible beside the final model?

Use your browser's downloads list to open northstar-cloud-decision.png.

Final Export Checklist

The image shows the Version 2: Hybrid Cloud panel.
The image shows the workload badges inside their chosen environments.
The image shows all five NIST cloud characteristics.
The image shows a characteristic connected to a business benefit.
The image shows the note NIST also defines Community cloud; Huawei's current fundamentals page focuses on Public, Private, and Hybrid cloud..

You have completed Northstar's cloud recommendation. The exported board now shows the failed assumption plus the workload decisions that repaired it.

Secret mission

Place a New Analytics Workload

Northstar's monthly analytics job crosses a control boundary that the earlier workloads did not. Extend the hybrid design so raw data stays local while approved summaries use a short public-cloud compute burst.

Clean Up Your Resources

Your draw.io decision board exists only in one editable source file plus two PNG exports. No paid cloud resources were created, so keeping these files has no ongoing cost.

Choose whether to keep the files, close the browser editor for now, or delete the files permanently.

Resources you used:

The editable source file northstar-cloud-decision.drawio.
The completed board export northstar-cloud-decision.png.
The analytics extension export northstar-cloud-decision-analytics.png.

Keep everything running

No action is needed. Choose this option if you want to revise the recommendation or show the finished decision board later.

Keep northstar-cloud-decision.drawio for future edits.
Keep northstar-cloud-decision.png as a snapshot of the original hybrid recommendation.
Keep northstar-cloud-decision-analytics.png as the final extended recommendation.

Pause - I'll come back to this later

Draw.io has no local process that needs to stay running for this board. Your saved files remain available after you close the browser editor.

Switch back to the draw.io browser editor from earlier.
Close the browser tab.
Keep all three project files together in their current local location.
Resume later by opening northstar-cloud-decision.drawio in draw.io.

Delete - I don't want to use this again

Remove all three local files if you no longer want to use this decision board.

Deletion Is Permanent

Deleting your work can feel drastic, so this option only fits when you no longer need the board. The PNG exports preserve images only.

The editable source is the file required for future revisions.

Use your computer's file manager to locate the three files listed above.
Select all three listed files.
Move the selected files to your system's deleted-items location.
Empty your system's deleted-items location.
Search for each listed filename in your file manager.

You should see no matching files. This confirms that the local project artifacts are gone.

Nice Work!

You did it! You turned Northstar Manufacturing's workload constraints into a defensible hybrid cloud recommendation.

Your decision board preserves the failed first attempt beside the improved design. That comparison makes your reasoning visible.

You've learned how to:

Build a one-page cloud decision board from a blank draw.io browser canvas. Retain the editable northstar-cloud-decision.drawio source beside two verified PNG exports.
Stress-test the Version 1 recommendation against Northstar's workload constraints. Expose the weak spots in its public cloud placement. Record a clear diagnosis of the one-size-fits-all assumption.
Map customer responsibilities across IaaS, PaaS, and SaaS. Use each control boundary to place workloads. Defend the final model with the five essential cloud characteristics.
Secret Mission: Extend the model with Monthly Factory Analytics. Keep Raw Sensor Data on premises. Route one-way Approved Summary Data to Processed Summary Analysis in the public cloud. Justify the analysis service with its control boundary plus two cloud characteristics.

Ready to quiz yourself?