Cloud · DevOps · Architecture
Architecture with the bill attached
Where the workload runs is an architecture decision and a financial one, and they are usually made by different people who never compare notes. This is cloud, colocation and on-premise work written with both in view — placement, migration, the scaffolding that has to exist first, and the diagrams that quietly stopped being true.
The right answer applied everywhere is the wrong answer — placement is a per-workload decision with a number attached.
Most migration regret traces to a decision made once and applied uniformly: everything to the cloud, or everything back. The writing here keeps landing on the same correction — the unit of decision is the workload, the evidence is the run-rate and the depreciation schedule, and the architecture diagram is only useful if somebody checks that the lines still describe what the system does.
The frame
Four questions before anything moves
Read the scaffolding piece →What is it, really
Steady-state or spiky, latency-bound or throughput-bound, regulated or not. The workload's own shape decides where it belongs, and the shape is measurable.
What does it cost where
Run-rate against capital, depreciation against a lease, and the egress nobody modelled. The comparison is arithmetic, not preference.
What has to exist first
Pipelines, environments, observability and the ability to roll back. Scaffolding built after the move is built during the incident.
What does the diagram claim
An architecture drawing is a set of assertions about the system. Most of them go stale silently, and the lines go wrong before the boxes do.
Flagships
What has to be true before you move anything
The Scaffolding Comes First
The environments, pipelines and rollback paths that have to exist before a migration is a migration rather than a hope. Written from the order these things actually get discovered in.
UML: The Boxes Were Right and the Lines Were Wrong
Architecture diagrams as assertions that can be checked, and what happens when nobody checks them: the boxes survive reorganisation and the relationships between them quietly stop being true.
Why it fits
Why placement is the through-line
Both sides of the decision
Held the architecture seat and the budget seat, frequently in the same meeting, which is where the disagreement between them becomes visible.
Data centre and cloud, not one camp
Colocation and owned hardware are not nostalgia and the cloud is not a default. The interesting work is knowing which workload belongs where and proving it.
Regulated placement
Data residency, PHI and the question of what leaves the building change the answer before cost does, and they are constraints these systems were built under.
The money is in the writing
Capital versus operating, depreciation schedules and vendor renewals appear throughout, because that is what the placement conversation is actually about.
Field notes
Cloud, DevOps & architecture writing
See all →- FedRAMP and GovRAMP: Evidence as a CommandGovernment Cloud Authorization
- Pareto in Practice: Eighty Percent from ThreeTesting & Build Engineering
- Engineering Team Turnarounds: Nobody Gets to Blame HistoryEngineering Leadership
- Local-First Medicine: The Box in the ClinicLocal-First Clinical AI
- Agile and Iterative Development: A Slice Through Every LayerSoftware Delivery & Engineering Practice
- What Building With AI Really Costs: The Whole BillAI Toolchain Costs & CI
- Self-Hosted CI Runners: The Job That Waited for a Sleeping PCCI/CD & Runner Infrastructure
- Root Cause Analysis: The Check Name GitHub Never RenderedRoot Cause Analysis
- Epic Migration Planning: Twelve Thousand Questions, One DateHealthcare IT Program Planning
- Hosted Code Review: The YAML Parsed and Nothing Read ItAI Code Review
- CapEx, OpEx, and EBITDA: The Same Dollar, Two Different StoriesTechnology Finance & Capital Strategy
- Depreciation and Useful Life: The Chip Outlives Its EconomicsInfrastructure Depreciation & Useful Life
- Placement Economics: The Right Answer, Applied EverywhereCloud, Colocation & Placement Economics
- Lift-and-Shift or Re-Architect: The Two That Modernized LaterCloud Migration Economics
- Vendor Economics: The Email in NovemberVendor Economics & Renewals
- AI Spend: Three Budgets Pretending to Be OneAI Cost & Procurement
- Data Pipelines for Machine Learning: The Duplicate That Hid the HeartbeatData Engineering & ML Pipelines
- Local-First Code Review: The Guard Dog and the LedgerLocal-First AI & Code Review
- Graph Engineering: The Edge That Carried NothingAgentic AI & Orchestration
- SAST and SCA: Nothing Matched the Patterns We Know AboutSecurity Scanning & Static Analysis
- Security Scanning in Practice: The Red Check Was a Full DiskSecurity Scanning & CI/CD Gates
- The AI Attack Surface: Four of the Six Layers Nobody ScansAI Security & Supply Chain
- UML: The Boxes Were Right and the Lines Were WrongModelling Notation & Diagrams
- Domain-Driven Design: The Word the Code Got WrongArchitecture & Domain Modelling
- The Scaffolding Comes FirstCI/CD & Delivery Pipelines
- What AI First Actually CostsAI Cost & Procurement
- Reasoning Without a ReasonerReasoning & Ontologies
- False GreenAI Code Review
- Reading Over the Machine's ShoulderAgentic AI
- The Zero Dollar AI Stack: Building a Full AI App Without Paying a CentTools
- The 6 Processors Powering Modern AI: CPU GPU TPU NPU LPU and DPU ExplainedConcepts
- The Pre-Existing Stack You Can’t IgnoreSecurity & Compliance
- MANY REPOS.MAX3
- CI_CD_CT_Product Evaluation – Qodo v. GeminiCode Reviews: Qodo & Gemini CA
- MAX3 · Technology EvaluationMAX3
- Four Models, One Repo, Zero GuaranteesCloud / DevOps
- DevOps in the Age of AICloud / DevOps
- Enhancing User Experience: The Power of Responsive and Reactive DesignCloud / DevOps
- How AWS Redesigns Legacy SaaS for Savings and SuccessCloud / DevOps
- Racing Toward the Future: The Accelerated Evolution of Machine LearningCloud / DevOps
- The Evolution of Software Development: From Specialized Experts to Near-Shore SolutionsCloud / DevOps
- Unlocking the Power of Cloud: Evolution and Benefits of Native ArchitectureCloud / DevOps
- Unveiling the Future: Harnessing the Power of Microservices in 2024Cloud / DevOps
Engage
Deciding where something should run?
Placement and migration review with the arithmetic shown — run-rate, depreciation, residency and the scaffolding that has to be in place before anything moves.
This site: HTTPS only · no third-party trackers · no analytics cookies · security.txt