6# Deprecation and Migration
7
8## Overview
9
10Code is a liability, not an asset. Every line of code has ongoing maintenance cost — bugs to fix, dependencies to update, security patches to apply, and new engineers to onboard. Deprecation is the discipline of removing code that no longer earns its keep, and migration is the process of moving users safely from the old to the new.
11
12Most engineering organizations are good at building things. Few are good at removing them. This skill addresses that gap.
13
14## When to Use
15
16- Replacing an old system, API, or library with a new one
17- Sunsetting a feature that's no longer needed
18- Consolidating duplicate implementations
19- Removing dead code that nobody owns but everybody depends on
20- Planning the lifecycle of a new system (deprecation planning starts at design time)
21- Deciding whether to maintain a legacy system or invest in migration
22
23## Core Principles
24
25### Code Is a Liability
26
27Every line of code has ongoing cost: it needs tests, documentation, security patches, dependency updates, and mental overhead for anyone working nearby. The value of code is the functionality it provides, not the code itself. When the same functionality can be provided with less code, less complexity, or better abstractions — the old code should go.
28
29### Hyrum's Law Makes Removal Hard
30
31With enough users, every observable behavior becomes depended on — including bugs, timing quirks, and undocumented side effects. This is why deprecation requires active migration, not just announcement. Users can't "just switch" when they depend on behaviors the replacement doesn't replicate.
32
33### Deprecation Planning Starts at Design Time
34
35When building something new, ask: "How would we remove this in 3 years?" Systems designed with clean interfaces, feature flags, and minimal surface area are easier to deprecate than systems that leak implementation details everywhere.
36
37## The Deprecation Decision
38
39Before deprecating anything, answer these questions:
40
41```
421. Does this system still provide unique value?
43 → If yes, maintain it. If no, proceed.
44
452. How many users/consumers depend on it?
46 → Quantify the migration scope.
47
483. Does a replacement exist?
49 → If no, build the replacement first. Don't deprecate without an alternative.
50
514. What's the migration cost for each consumer?
52 → If trivially automated, do it. If manual and high-effort, weigh against maintenance cost.
53
545. What's the ongoing maintenance cost of NOT deprecating?
55 → Security risk, engineer time, opportunity cost of complexity.
56```
57
58## Compulsory vs Advisory Deprecation
59
60| Type | When to Use | Mechanism |
61|------|-------------|-----------|
62| **Advisory** | Migration is optional, old system is stable | Warnings, documentation, nudges. Users migrate on their own timeline. |
63| **Compulsory** | Old system has security issues, blocks progress, or maintenance cost is unsustainable | Hard deadline. Old system will be removed by date X. Provide migration tooling. |
64
65**Default to advisory.** Use compulsory only when the maintenance cost or risk justifies forcing migration. Compulsory deprecation requires providing migration tooling, documentation, and support — you can't just announce a deadline.
66
67## The Migration Process
68
69### Step 1: Build the Replacement
70
71Don't deprecate without a working alternative. The replacement must:
72
73- Cover all critical use cases of the old system
74- Have documentation and migration guides
75- Be proven in production (not just "theoretically better")
76
77### Step 2: Announce and Document
78
79```markdown
80## Deprecation Notice: OldService
81
82**Status:** Deprecated as of 2025-03-01
83**Replacement:** NewService (see migration guide below)
84**Removal date:** Advisory — no hard deadline yet
85**Reason:** OldService requires manual scaling and lacks observability.
86 NewService handles both automatically.
87
88### Migration Guide
891. Replace import { client } from 'old-service' with import { client } from 'new-service'
902. Update configuration (see examples below)
913. Run the migration verification script: npx migrate-check
92```
93
94### Step 3: Migrate Incrementally
95
96Migrate consumers one at a time, not all at once. For each consumer:
97
98```
991. Identify all touchpoints with the deprecated system
1002. Update to use the replacement
1013. Verify behavior matches (tests, integration checks)
1024. Remove references to the old system
1035. Confirm no regressions
104```
105
106**The Churn Rule:** If you own the infrastructure being deprecated, you are responsible for migrating your users — or providing backward-compatible updates that require no migration. Don't announce deprecation and leave users to figure it out.
107
108### Step 4: Remove the Old System
109
110Only after all consumers have migrated:
111
112```
1131. Verify zero active usage (metrics, logs, dependency analysis)
1142. Remove the code
1153. Remove associated tests, documentation, and configuration
1164. Remove the deprecation notices
1175. Celebrate — removing code is an achievement
118```
119
120## Migration Patterns
121
122### Strangler Pattern
123
124Run old and new systems in parallel. Route traffic incrementally from old to new. When the old system handles 0% of traffic, remove it.
125
126```
127Phase 1: New system handles 0%, old handles 100%
128Phase 2: New system handles 10% (canary)
129Phase 3: New system handles 50%
130Phase 4: New system handles 100%, old system idle
131Phase 5: Remove old system
132```
133
134### Adapter Pattern
135
136Create an adapter that translates calls from the old interface to the new implementation. Consumers keep using the old interface while you migrate the backend.
137
138```typescript
139// Adapter: old interface, new implementation
140class LegacyTaskService implements OldTaskAPI {
141 constructor(private newService: NewTaskService) {}
142
143 // Old method signature, delegates to new implementation
144 getTask(id: number): OldTask {
145 const task = this.newService.findById(String(id));
146 return this.toOldFormat(task);
147 }
148}
149```
150
151### Feature Flag Migration
152
153Use feature flags to switch consumers from old to new system one at a time:
154
155```typescript
156function getTaskService(userId: string): TaskService {
157 if (featureFlags.isEnabled('new-task-service', { userId })) {
158 return new NewTaskService();
159 }
160 return new LegacyTaskService();
161}
162```
163
164### Database Schema Migrations (Expand/Contract)
165
166A schema change is the riskiest migration because the data is the one thing you cannot roll back by reverting a deploy. The failure mode is coupling the schema change to the code change: rename a column in the same release that starts using the new name, and during the rollout window — when old and new code run at once — one of them is querying a column that doesn't exist. The fix is to **never change a column in place**. Migrate in additive phases so old and new code are both valid at every step.
167
168```
169EXPAND ──────────────→ MIGRATE ──────────────→ CONTRACT
170add the new column, backfill existing rows, once no code reads the
171nullable, alongside dual-write old+new from old column, drop it in
172the old one the app a later, separate deploy
173```
174
175**Worked example — renaming name to full_name:**
176
1771. **Expand.** Add full_name as nullable. Deploy. (Old code ignores it; nothing breaks.)
1782. **Dual-write.** App writes both name and full_name on every insert/update. Deploy.
1793. **Backfill.** Copy name → full_name for existing rows, in batches, so you don't lock the table.
1804. **Switch reads.** Point the app at full_name, keep writing both. Deploy and bake.
1815. **Contract.** Stop writing name, then — in a *separate, later* deploy — drop the column.
182
183Each step is independently deployable and reversible: if step 4 misbehaves, roll the code back and full_name is still being populated. Treat each phase as a thin vertical slice — see the incremental-implementation skill.
184
185**Rules:**
186- **Additive first, destructive last and alone.** Adds (new nullable column, new table, new index) are safe in any deploy; drops and renames get their own deploy *after* no code references the old shape.
187- **Every migration has a tested down path.** A migration you can't reverse is a deploy you can't roll back. Write and run the down before merging.
188- **Backfill in batches, off the hot path.** A single UPDATE over millions of rows locks the table; chunk it and throttle.
189- **Build large indexes without blocking writes** (e.g. Postgres CREATE INDEX CONCURRENTLY).
190- **Decouple from code by feature flag** when the cutover is risky, exactly as in the Feature Flag Migration pattern above.
191
192## Zombie Code
193
194Zombie code is code that nobody owns but everybody depends on. It's not actively maintained, has no clear owner, and accumulates security vulnerabilities and compatibility issues. Signs:
195
196- No commits in 6+ months but active consumers exist
197- No assigned maintainer or team
198- Failing tests that nobody fixes
199- Dependencies with known vulnerabilities that nobody updates
200- Documentation that references systems that no longer exist
201
202**Response:** Either assign an owner and maintain it properly, or deprecate it with a concrete migration plan. Zombie code cannot stay in limbo — it either gets investment or removal.
203
204## Common Rationalizations
205
206| Rationalization | Reality |
207|---|---|
208| "It still works, why remove it?" | Working code that nobody maintains accumulates security debt and complexity. Maintenance cost grows silently. |
209| "Someone might need it later" | If it's needed later, it can be rebuilt. Keeping unused code "just in case" costs more than rebuilding. |
210| "The migration is too expensive" | Compare migration cost to ongoing maintenance cost over 2-3 years. Migration is usually cheaper long-term. |
211| "We'll deprecate it after we finish the new system" | Deprecation planning starts at design time. By the time the new system is done, you'll have new priorities. Plan now. |
212| "Users will migrate on their own" | They won't. Provide tooling, documentation, and incentives — or do the migration yourself (the Churn Rule). |
213| "We can maintain both systems indefinitely" | Two systems doing the same thing is double the maintenance, testing, documentation, and onboarding cost. |
214| "Just rename the column, it's one line" | During the rollout, old and new code run together — one will query a column that no longer exists. Expand/contract, never rename in place. |
215| "I'll add the column and drop the old one in the same migration" | That couples a safe add to a destructive drop. Drops get their own deploy, after no code references the old shape. |
216| "We'll write the rollback if we need it" | A migration with no down path is a deploy you can't reverse. Write and run the down before merging. |
217
218## Red Flags
219
220- Deprecated systems with no replacement available
221- Deprecation announcements with no migration tooling or documentation
222- "Soft" deprecation that's been advisory for years with no progress
223- Zombie code with no owner and active consumers
224- New features added to a deprecated system (invest in the replacement instead)
225- Deprecation without measuring current usage
226- Removing code without verifying zero active consumers
227- A schema change and the code that depends on it shipped in the same deploy
228- A column renamed or dropped in place rather than via expand/contract
229- A migration merged with no tested down path, or a backfill that locks the table
230
231## Verification
232
233After completing a deprecation:
234
235- [ ] Replacement is production-proven and covers all critical use cases
236- [ ] Migration guide exists with concrete steps and examples
237- [ ] All active consumers have been migrated (verified by metrics/logs)
238- [ ] Old code, tests, documentation, and configuration are fully removed
239- [ ] No references to the deprecated system remain in the codebase
240- [ ] Deprecation notices are removed (they served their purpose)
241
242After a database schema migration:
243
244- [ ] The change ships in additive phases (expand → backfill → contract), not a single in-place edit
245- [ ] Old and new code are both valid against the schema at every deploy step
246- [ ] Each migration has a tested down path; backfills run in throttled batches
247- [ ] Destructive steps (drop/rename) ship in their own deploy after no code references the old shape
248