Skip to content
Ameen Abozaid
Englishالعربية
ARC-023

Data architecture

  • 6 Lessons
  • 3 h practice (estimate)
Start module
On this page 9 topics

Overview

Topics

  • Operational vs analytical workloads
  • data lifecycle

Learn first

3 h of practice (estimate)

Learning outcomes

  1. Separate operational (transactional) and analytical workloads and explain why they need different stores.
  2. Describe the data lifecycle (creation, use, sharing, archiving, deletion) for each major data set.
  3. Choose between a data warehouse, a data lake, and a lakehouse for a reporting need.
  4. Produce a conceptual data model and a data-flow view for a solution.

Lessons

  1. 01Data architecture for solution architects
  2. 02Operational vs analytical workloads
  3. 03Warehouses, lakes, and lakehouses
  4. 04The data lifecycle
  5. 05Conceptual models and data-flow views
  6. 06Demo: Larkspur's data landscape from sales floor to board report

Lab and exercises

Lab

ARC-023-LAB1 · Larkspur data architecture

Outcomes: 1–4 · Time: ~3 h

  1. List every Larkspur data set from the case brief and interview notes.
  2. Draw the conceptual data model and mark one system of record per entity.
  3. Draw the data-flow view from stores and web to ERP, CRM, platform, and reporting; label each arrow with data set and trigger.
  4. Choose warehouse, lake, or lakehouse for the three reporting needs and justify.
  5. Fill the lifecycle table: stages, retention with its source, copies, deletion trigger.
  6. Defend the view to a role-played head of reporting.

Acceptance: every data set has one system of record and a defined retention.

Exercises (case-based; six exercises)

IDOutcomeTask
ARC-023-EX011The store-ops lead runs a nightly stock report on the production order database and checkout slows at 02:00. Write a half-page note explaining the cause and the target design.
ARC-023-EX022For "customer profile", fill a five-stage lifecycle with retention, copies, and deletion trigger.
ARC-023-EX033Choose warehouse, lake, or lakehouse for: finance daily sales, store camera footfall counts, demand-forecast training. Justify each in two lines.
ARC-023-EX044Draw a conceptual model for returns (Customer, Order, ReturnRequest, Refund, Product) and mark systems of record.
ARC-023-EX051, 4Two systems both claim to own product prices. Propose the system of record, the flow, and what the other system may do.
ARC-023-EX062, 3Review a data architecture with no deletion stage and a lake holding copies of customer files; give three corrections.
Show solutions
  • EX01 Long scans compete with checkout for I/O and locks; move reporting to a replica or warehouse fed by change capture; keep the store report on the analytical copy.
  • EX02 Create at sign-up (CRM and platform); use while the account is active; share with the delivery carrier (address only); archive after inactivity period set by policy; delete on request or at the retention end, including warehouse and backups; verify periods with legal.
  • EX03 Finance: warehouse (modeled, reconciled); footfall: lake (raw, semi-structured, exploratory); forecast training: lakehouse or lake with a curated layer, depending on team skills.
  • EX04 Customer, Order, ReturnRequest (platform), Refund (payment provider and ERP posting), Product (ERP); relations Order 1-many ReturnRequest, ReturnRequest 1-1 Refund.
  • EX05 ERP is system of record; platform consumes prices through an interface and caches; platform may propose promotions through ERP, not edit prices locally.
  • EX06 Add deletion and archive stages; list all copies with owners; apply the same retention to the lake as to the source; add classification to lake folders.

All four outcomes are also covered by the lab acceptance and the quiz.

Workplace evidence: a data architecture view you produced that guided a reporting or platform decision.

Quiz

Outcomes: O1 operational vs analytical · O2 data lifecycle · O3 warehouse, lake, lakehouse · O4 conceptual model and data flow.

A self-check for you: nothing is graded, saved or sent anywhere. Choose an answer, then check it. You can try again as often as you like.

  1. 1A weekly margin report scans three years of orders and slows the web shop. The best fix is:O1
  2. 2Which workload pattern is typical of OLTP?O1
  3. 3Which lifecycle stage do teams most often forget?O2
  4. 4Why record all copies of a data set?O2
  5. 5Which store best fits finance-signed KPI reporting with modeled, reconciled data?O3
  6. 6What does a lakehouse add to a lake?O3
  7. 7A conceptual data model shows:O4
  8. 8On a data-flow view, an arrow into a system that is not the system of record represents:O4
  9. 9Your data architecture review finds reporting running on the production database. Which common mistake is this?O1
  10. 10Where should the retention period for a data set come from?O2

Practical part

Data architecture review of the Larkspur lab: the reviewer checks one system of record and a defined retention for every data set. Pass: quiz at least 8/10, at least five of six exercises correct, and a lab with no data set missing an owner or retention.

Common mistakes

Reporting directly off production databases; data sets with no system of record; ignoring deletion.

Assessment

Data architecture review.

Workplace evidence

A data architecture view you produced that guided a reporting or platform decision.

Career relevance

The data section of every solution architecture pack.

Type to search levels, phases, modules and lessons.