← कोर्सच्या मुख्य पानाकडे परत

⏮️ आधी काय होते & फायदे-तोटे

प्रत्येक साधनाने काहीतरी वाईट गोष्ट बदलली — आणि तेच साधन कुठेतरी चुकीचे ठरते. प्रत्येक मोठ्या कल्पनेसाठी: आधी जीवन कसे होते, प्रामाणिक फायदे ✅ / तोटे ❌, आणि कुठे वापरावी 👍 विरुद्ध कुठे नाही 👎.

🧭 Tickets vs self-service — lessons 01, 05

⏮️ Before platforms

Every request went to an operations team's queue and was done by hand, first come, first served.

✅ फायदे

  • self-service: minutes instead of days; the same result every time
  • tickets: a person can judge an unusual request

❌ तोटे

  • self-service: the office must build and maintain the menu, quotas and cleanup
  • tickets: waits grow whenever requests outpace the team; every result is slightly different

👍 वापरा जेव्हा

  • self-service for common, repeat requests (databases, buckets, repos, DNS)
  • tickets for rare, risky or truly unusual requests

👎 दोनदा विचार करा जेव्हा

  • self-service with no quotas or delete path
  • tickets for the tenth identical database this month

🤝 Mandated vs earned adoption — lesson 02

⏮️ Before platform-as-a-product

A central team chose a tool and ordered everyone to use it by a date.

✅ फायदे

  • earned: adoption is honest feedback; teams that leave tell you what to fix
  • mandate: fast, uniform, sometimes needed for compliance

❌ तोटे

  • earned: slower; some teams stay off the path for a long time
  • mandate: usage says nothing about quality; workarounds and resentment grow

👍 वापरा जेव्हा

  • earn adoption for developer tools and workflows
  • mandate only real non-negotiables (security, legal), and explain why

👎 दोनदा विचार करा जेव्हा

  • mandating a tool nobody was interviewed about
  • a 'product' with no support channel, docs or roadmap

🛤️ A golden-path template vs a blank page — lesson 03

⏮️ Before templates

Each new service started from an empty repo, or a copy of whichever repo was nearest.

✅ फायदे

  • template: good defaults (tests, limits, pipeline, owner) from minute one; services look alike
  • blank page: total freedom for an unusual service

❌ तोटे

  • template: templates rot without an owner; day-2 updates need their own plan
  • blank page: every team re-solves the same problems, differently

👍 वापरा जेव्हा

  • templates for the common service shapes, owned and versioned
  • leaving the path when a service truly differs — and owning the result

👎 दोनदा विचार करा जेव्हा

  • a template with every option exposed
  • copying an old repo because the template is harder to use

🚚 Moving tag vs exact pin vs pasted copy — lesson 07

⏮️ Before shared pipelines

Every repo had its own copied pipeline file, edited by hand over the years.

✅ फायदे

  • moving tag (@v2): fixes arrive by themselves
  • exact pin (@v2.0.0 or a SHA): nothing changes until you choose

❌ तोटे

  • moving tag: a bad release reaches everyone at once
  • exact pin: falls behind unless someone bumps it; a copy never gets any fix

👍 वापरा जेव्हा

  • moving major tags for the office's own well-tested templates
  • exact SHAs for third-party code, with automated bump PRs

👎 दोनदा विचार करा जेव्हा

  • pasted copies of the shared pipeline
  • moving a major tag to a breaking release

🛡️ Gates vs guardrails — lesson 09

⏮️ Before policy as code

A person reviewed changes against a written checklist, or nobody checked at all.

✅ फायदे

  • guardrail: automatic, consistent, runs on every change; says how to fix
  • gate: a person can weigh context

❌ तोटे

  • guardrail: rules need owners, tests and an exception path
  • gate: slow, inconsistent, becomes a queue of its own

👍 वापरा जेव्हा

  • guardrails for clear rules (owner, pinned images, limits, private buckets), audit first
  • a human review for rare, high-risk changes

👎 दोनदा विचार करा जेव्हा

  • enforce on day one with no audit period
  • rules that only say 'denied'

📊 Metrics as signals vs as targets — lesson 11

⏮️ Before measuring

The platform's value was judged by feeling, or by how many tickets it closed.

✅ फायदे

  • signals: DORA keys show a team's trend; SPACE adds what people say
  • targets: simple to announce

❌ तोटे

  • signals: need a baseline and patience; small samples mislead
  • targets: Goodhart's law — the number moves, the work does not

👍 वापरा जेव्हा

  • each team's own trend, before vs after the paved path, next to survey answers
  • time to first deploy for new teams

👎 दोनदा विचार करा जेव्हा

  • ranking teams by deployment frequency
  • measuring activity (commits, lines) as productivity