CSDV3017  ·  DEVOPS  ·  SCHOOL OF COMPUTER SCIENCE, UPES

Metrics Tools, DevOps Lifecycle & Digital Transformation

Lecture 12 — measuring what matters with metrics dashboards, mapping the full DevOps lifecycle end-to-end, and understanding how DevOps powers digital transformation in modern enterprises.

InstructorDr. Mohsin Furkh Dar
SessionWeek 4 · Tue, 30 Jun 2026
Time14:00 – 15:00
UnitUnit IV
I
II
III
IV
V
VI
VII
WHERE WE LEFT OFF

Recap & today's agenda

Lecture 11 · Done

IaC, CD & Jenkins

  • Infrastructure as Code — Terraform, Ansible, Puppet
  • Continuous Delivery (manual gate) vs Deployment (fully automated)
  • Deployment strategies: Blue-Green, Canary, Rolling
  • Jenkins — Controller/Agent, Jenkinsfile, pipeline stages
Lecture 12 · Today

Metrics, Lifecycle & DX

  • Metrics tools — Prometheus, Grafana, ELK, Datadog
  • The DevOps lifecycle — the complete 8-phase infinity loop
  • Digital transformation — why DevOps is the engine behind it
  • Real-world case studies — Netflix, Amazon, Target
MEASURE TO IMPROVE

If you can't measure it, you can't improve it

"Without data, you're just another person with an opinion." — W. Edwards Deming

Question 01

How fast do we ship?

Deployment Frequency and Lead Time for Changes tell you how quickly ideas become running code in production.

Question 02

How reliable are we?

Change Failure Rate and MTTR tell you how often deployments break things and how quickly your team recovers.

Question 03

Where's the bottleneck?

Cycle time analytics reveal which pipeline stages are slow. Optimize the bottleneck and the whole system speeds up.

METRICS COLLECTION

Prometheus — the metrics engine

What Is It?

Open-Source Metrics Database

  • Created at SoundCloud in 2012; now a CNCF graduated project
  • Pull-based model: Prometheus scrapes metrics from your apps every 15s
  • Stores data as time-series (metric name + labels + timestamp + value)
  • Built-in query language: PromQL
  • Alerting rules trigger when thresholds are breached
Example

What a metric looks like

# A Prometheus metric exposed by your app

http_requests_total{
  method="GET",
  endpoint="/api/users",
  status="200"
} 14523

# PromQL: requests per second over 5 min
rate(http_requests_total[5m])
VISUALIZATION

Grafana — making metrics beautiful

Overview

The Dashboard Tool

  • Open-source visualization platform; the "face" of Prometheus data
  • Connects to 50+ data sources (Prometheus, Elasticsearch, PostgreSQL, CloudWatch)
  • Drag-and-drop dashboard builder with real-time auto-refresh
  • Pre-built templates: Node Exporter (CPU/RAM), Kubernetes cluster, JVM, Nginx
Example Dashboard

What a Grafana panel shows

  • Line chart: Request latency (p50, p95, p99) over 24 hours
  • Gauge: Current CPU utilization — 67%
  • Stat panel: Uptime: 99.97%
  • Heatmap: Error rate distribution by hour of day
  • Alert annotation: Red vertical line when an incident was detected

The Prometheus + Grafana stack is the industry standard for open-source monitoring. Prometheus collects and stores the data; Grafana queries PromQL and renders it as interactive dashboards.

COMPLEMENTARY TOOLS

ELK Stack & Datadog

Open-Source

ELK Stack (Logs)

  • Elasticsearch: Distributed search engine — indexes and stores log data
  • Logstash / Fluentd: Collects, parses, and ships logs from all servers
  • Kibana: Web UI to search, filter, and visualize logs
  • Example: Search all logs containing "NullPointerException" in the last 30 minutes, across 50 microservices, in one query
Commercial / SaaS

Datadog

  • All-in-one: metrics + logs + traces + security in a single platform
  • 600+ integrations (AWS, K8s, Docker, Jenkins, GitHub)
  • AI-powered anomaly detection — spots unusual patterns automatically
  • Example: Datadog detects that p99 latency spiked 300% after the latest deploy and auto-creates an incident in PagerDuty
TOOL LANDSCAPE

Metrics tool comparison

Tool Type Best For Cost
Prometheus Metrics DB Time-series metrics collection; alerting via AlertManager Free (open-source)
Grafana Visualization Dashboards for Prometheus, Elasticsearch, any data source Free (OSS) / Cloud paid tier
ELK Stack Log Management Centralized log search, analysis, and visualization Free (OSS) / Elastic Cloud paid
Datadog All-in-One SaaS Unified metrics + logs + traces + security $15–$34 per host/month
New Relic APM SaaS Application performance monitoring, distributed tracing Free tier / $0.30 per GB ingested
PagerDuty Incident Mgmt On-call scheduling, alert routing, escalation policies $21–$41 per user/month
THE BIG PICTURE

The DevOps Lifecycle — 8 phases

Plan
→
Code
→
Build
→
Test
→
Release
→
Deploy
→
Operate
→
Monitor
Dev Loop (Left)

Plan → Code → Build → Test

  • Plan: User stories, sprint planning (Jira, Azure Boards)
  • Code: Write features, branch in Git, open PRs
  • Build: Compile code, resolve dependencies (Maven, npm)
  • Test: Unit, integration, E2E tests run automatically
Ops Loop (Right)

Release → Deploy → Operate → Monitor

  • Release: Package artifact, tag version, approve for deployment
  • Deploy: Push to production via Blue-Green / Canary / Rolling
  • Operate: Manage infrastructure, scale, handle incidents
  • Monitor: Observe metrics, logs, traces → feed data back to Plan

The ∞ shape matters: Monitor feeds back into Plan. The loop never ends — every cycle makes the product better.

TOOLCHAIN MAP

One tool per lifecycle phase

Phase Activity Example Tools
Plan Requirements, user stories, sprint planning Jira, Azure Boards, Trello, Linear
Code Write code, collaborate via PRs VS Code, Git, GitHub, GitLab
Build Compile, resolve dependencies, package Maven, Gradle, npm, Docker
Test Automated unit, integration, E2E testing JUnit, pytest, Selenium, Cypress
Release Version tagging, approval gates GitHub Releases, Jenkins, GitLab CI
Deploy Push to staging/production environments Terraform, Ansible, Kubernetes, ArgoCD
Operate Manage infrastructure, auto-scale AWS, Azure, GCP, Kubernetes, Puppet
Monitor Observe, alert, investigate Prometheus, Grafana, ELK, Datadog, PagerDuty
DIGITAL TRANSFORMATION

What is Digital Transformation?

Definition

Reimagining business through technology

Digital Transformation (DX) is the process of using digital technology to fundamentally change how an organization operates, delivers value to customers, and competes in the market. It's not just "going digital" — it's rethinking business models, internal processes, and customer experiences from the ground up.

$3.9T
Global DX spending by 2027

Source: IDC, 2024

70%
of DX initiatives fail

Usually due to culture, not technology

2.5×
Revenue growth for DX leaders

vs. laggards (McKinsey, 2023)

FOUR PILLARS

The 4 pillars of Digital Transformation

Pillar 01

Customer Experience

Personalization, omnichannel experiences, mobile-first design. Example: Netflix's AI-driven recommendations.

Pillar 02

Operational Agility

Automating internal workflows, reducing manual handoffs. Example: Amazon's automated warehouses and one-click ordering.

Pillar 03

Workforce Enablement

Cloud collaboration tools, self-service platforms, upskilling teams. Example: Microsoft Teams replacing email-first culture.

Pillar 04

Digital Tech Integration

Cloud, AI/ML, IoT, Big Data, microservices. Example: Tesla's over-the-air software updates to millions of cars.

THE CONNECTION

DevOps is the engine of Digital Transformation

"You can't have digital transformation without DevOps. The speed of business depends on the speed of software." — Gene Kim, The Phoenix Project

How DevOps Enables DX

The multiplier effect

  • Speed: CI/CD lets teams deploy features in hours, not months
  • Reliability: Automated testing and monitoring reduce risk
  • Scalability: IaC + containers spin up capacity on demand
  • Experimentation: Feature flags + canary deploys enable A/B testing safely
  • Feedback: Monitoring data feeds directly into product decisions
Without DevOps

DX bottleneck

  • Marketing promises a new feature → IT says "6 months"
  • Quarterly releases mean competitors ship first
  • Manual deployments cause weekend outages
  • No monitoring → customer complaints are the alerting system
  • "Transformation" becomes just a PowerPoint presentation
CASE STUDY

Netflix — DevOps at planetary scale

Before DevOps (2007)

The old Netflix

  • Monolithic Java application running in its own data center
  • A single database corruption caused a 3-day outage
  • DVD mailing was still the primary business model
  • Deployments happened every few weeks with major downtime
After DevOps (Today)

The Netflix machine

  • 1,000+ microservices running on AWS
  • Thousands of deployments per day with zero downtime
  • Chaos Monkey — intentionally kills production servers to test resilience
  • 238 million subscribers in 190 countries

Key lesson: Netflix didn't just adopt DevOps tools — they reimagined their entire culture and architecture. The "Freedom and Responsibility" culture empowers any engineer to deploy to production without approval, because automated testing and monitoring catch problems instantly.

MORE EXAMPLES

Amazon & Target — two ends of the spectrum

Born-Digital Giant

Amazon

  • Deploys code every 11.7 seconds on average
  • Built AWS (the cloud itself) to solve their own scaling problems
  • "Two-pizza teams" — small, autonomous squads that own their service end-to-end
  • Every team runs its own CI/CD pipeline and on-call rotation

Result: From bookstore to $1.5 trillion tech conglomerate.

Legacy Retailer Transformed

Target

  • 2013 data breach exposed 40M credit card records — a DX wake-up call
  • Moved from outsourced IT to 4,000 in-house engineers
  • Adopted microservices, Kubernetes, and CI/CD pipelines internally
  • Built custom tools: Drone (CI) and TAP (internal PaaS)

Result: E-commerce revenue grew 195% (2019-2022).

ORGANIZATIONAL ROLE

Where DevOps sits in a modern organization

Key Roles

DevOps is not a single team — it's a culture

In mature organizations, DevOps isn't a silo; it's a shared responsibility. However, specific roles exist to enable it:

Role Responsibility Tools They Own
DevOps Engineer Build and maintain CI/CD pipelines, IaC, and developer tooling Jenkins, GitHub Actions, Terraform
SRE (Site Reliability) Define SLOs, manage incident response, reduce toil through automation Prometheus, PagerDuty, runbooks
Platform Engineer Build the Internal Developer Platform (IDP) that abstracts infrastructure Kubernetes, Backstage, ArgoCD
Cloud Architect Design cloud infrastructure, networking, security, cost optimization AWS, Azure, GCP console, Terraform
WRAP-UP

Summary & what's next

Lecture 12 · Key Takeaways

What you should remember

  • Metrics Tools: Prometheus (collection) + Grafana (visualization); ELK (logs); Datadog (all-in-one SaaS)
  • DevOps Lifecycle: 8 phases in an infinity loop — Plan → Code → Build → Test → Release → Deploy → Operate → Monitor → back to Plan
  • Digital Transformation: Reimagining business through technology; 4 pillars: CX, Operational Agility, Workforce, Tech Integration
  • DevOps as DX Engine: Speed, reliability, scalability, experimentation, and feedback are what make DX real — not just PowerPoint
  • Case Studies: Netflix (monolith → 1,000 microservices), Amazon (deploy every 11.7s), Target (legacy → modern engineering)
Next Lecture · Lecture 13

Mon, 06 Jul 2026 · 14:00–15:00 · Unit V

Selecting right DevOps tools; Docker & Kubernetes; Puppet & Ansible; Monitoring tools; IBM Case Study: CI/CD.

Unit IV Complete ✓

You've finished Unit IV!

Lectures 10-12 covered DevOps principles, Git, Gitflow, CI, IaC, CD, Jenkins, metrics, lifecycle, and DX. Next: hands-on tooling in Unit V.

CSDV3017 · DEVOPS
SHEET 01/16