Testar grátis
4 min leitura Guide 374 of 877

How to Manage Technical Spikes in Agile Development?

How to manage technical spikes in agile development?

Manage technical spikes by creating time-boxed research tasks with clear questions to answer, documenting findings in NoteVault or task comments, and converting outcomes to actionable development tasks. Use spike: labels, set maximum duration (1-5 days), and require a decision document as the definition of done.

Spike task template

## Spike: Evaluate GraphQL vs REST for new API

Time-box: 3 days max
Due: 2025-01-30

Questions to answer:
1. Can GraphQL handle our query patterns?
2. What's the learning curve for the team?
3. How does caching work?
4. What's the performance difference?

Evaluation criteria:
- Team familiarity
- Performance requirements
- Client needs (mobile, web)
- Tooling ecosystem

Definition of Done:
- [ ] Questions answered
- [ ] Findings documented
- [ ] Recommendation made
- [ ] Follow-up tasks created

Spike labels

LabelPurpose
spike:researchInvestigation task
spike:pocProof of concept
spike:evaluationComparing options
spike:1-day1 day time-box
spike:3-day3 day time-box
spike:5-day5 day time-box

Spike types and time-boxes

Spike TypeDurationPurpose
Quick investigation1 daySimple question
Technology evaluation3 daysCompare options
Proof of concept3-5 daysBuild prototype
Architecture spike5 daysDesign decision
Performance spike2-3 daysMeasure approach

Spike workflow:

  1. Identify uncertainty - What do we need to learn?
  2. Define questions - Specific, answerable questions
  3. Set time-box - Maximum investigation time
  4. Create spike task - Use template
  5. Investigate - Research, prototype, test
  6. Document findings - NoteVault or task comments
  7. Make recommendation - Based on evidence
  8. Create follow-up - Development tasks from spike
  9. Close spike - Time-box ends, document state

Spike documentation template

# Spike: [Title] - Findings

## Summary
Brief recommendation in 1-2 sentences.

## Questions & Answers

### Q1: Can GraphQL handle our query patterns?
**Answer**: Yes, with caveats. [Evidence and details]

### Q2: What's the learning curve?
**Answer**: 2-3 sprints for proficiency. [Evidence]

## Recommendation
Recommend GraphQL for new API because...

## Estimated effort
- Initial setup: 1 sprint
- Team training: 1 sprint
- Migration: 3 sprints

## Risks
- Learning curve may slow initial velocity
- Caching more complex than REST

## Follow-up Tasks
- [ ] Create GraphQL schema design task
- [ ] Create training session task
- [ ] Update architecture docs

Common spike mistakes

MistakePrevention
No time-boxAlways set maximum duration
Undefined questionsWrite specific questions
No documentationRequire findings doc
Endless spikeHard stop at time-box
No follow-upCreate tasks from findings