The deadline is close: do you ship a quick fix or take longer to build a cleaner solution? A tradeoff means gaining a benefit while accepting a cost, risk, or limitation elsewhere. Neither speed nor long-term quality wins every time; the goal is to pick your pain on purpose. section 1.2 section 1.10
Good choices depend on the business need, not just technical elegance. Ask why the feature matters, what constraints are real, and which qualities matter most. A shortcut can be reasonable if its debt is acknowledged and there is a plan to address it; making shortcuts an unexamined habit is another story. section 1.4 section 1.2 section 1.5.3
A practical way to make a defensible decision:
- Clarify the request, its purpose, and the constraints before choosing a design. section 2.1
- Sketch the idea and identify at least two realistic options. List a benefit and a cost or risk for each, tied to the priorities you uncovered. section 1.5.5 section 1.8.1
- Record the choice, its rationale, and its consequences. Where possible, track results such as incident rates or feature delivery time. section 1.10 section 2.2
Shipping favorites
Adding favorites to an existing user profile is quick and simple, but counting favorites becomes expensive as users grow. A concurrent save can also overwrite another favorite. Dedicated storage improves scalability and maintainability, but takes longer to ship and adds query and infrastructure complexity. If a functional first version is needed in two weeks, shipping speed may take priority—provided the risks are understood. section 1.5.5 section 1.5.3
A faster dashboard
A cache can avoid repeated slow database queries, but it adds a component to deploy and monitor. It can also leave the dashboard up to 10 minutes out of date. The decision is not simply “faster is better”: ask whether that delay is acceptable for the people using the dashboard. section 2.4.4
Tradeoffs also shape system-wide choices. A monolith uses in-memory calls but can couple components and deployments; microservices can scale independently but must coordinate across a network. A modular monolith keeps a single deployment with stricter internal boundaries. None is automatically the hero. section 12.1 section 12.2.6
Table 12.1 Architecture Spectrum Table
|
Feature
|
The Monolith
|
The Microservice
|
The Modular Monolith
|
|
Deployment Unit
|
Single executable
|
Multiple independent services
|
Single executable
|
|
Component Interaction
|
In-memory function calls
|
Network calls (REST/Events)
|
In-memory via strict interfaces
|
|
Scaling
|
All-Or-Nothing (Vertical)
|
Independent (Horizontal)
|
All-Or-Nothing (Vertical)
|
|
Data Storage
|
Single shared database
|
Database-per-service
|
Single database (logically separated)
|
|
Advantage
|
Ultimate simplicity and zero latency
|
Extreme agility and independent scaling
|
Clean boundaries without the network tax
|
|
Risk
|
High coupling and deployment jams
|
Distributed data complexity
|
Requires strict developer discipline
|
Compare deployment, communication, scaling, and risks across the three architecture styles. section 12.2.6
There is no perfect architecture. Choose the option whose benefits fit the current goal and whose downsides the team can accept, then make those downsides visible. section 2.2 section 1.5.5










