Govardhana Miriyala Kannaiah

Cloud Cost Reduction Techniques 👇

Each technique enables organizations to optimize cloud usage and minimize expenses

𝟭. 𝗥𝗲𝗱𝘂𝗰𝗲 𝗨𝘀𝗮𝗴𝗲:
Optimize resource quantity and size while maintaining application performance (eg:- Trim instances, shrink storage, streamline services)

𝟮. 𝗧𝗲𝗿𝗺𝗶𝗻𝗮𝘁𝗲 𝗜𝗱𝗹𝗲 𝗥𝗲𝘀𝗼𝘂𝗿𝗰𝗲𝘀: 
Identify and terminate any idle or unused resources such as instances, databases, or storage volumes

𝟯. 𝗥𝗶𝗴𝗵𝘁 𝗦𝗶𝘇𝗶𝗻𝗴:
Right-size instances to match application needs of your applications

𝟰. 𝗦𝗵𝘂𝘁𝗱𝗼𝘄𝗻 𝗥𝗲𝘀𝗼𝘂𝗿𝗰𝗲𝘀 𝗗𝘂𝗿𝗶𝗻𝗴 𝗜𝗻𝗮𝗰𝘁𝗶𝘃𝗲 𝗣𝗲𝗿𝗶𝗼𝗱𝘀:
Implement automated processes or schedules to shut down non-critical resources during inactive periods

𝟱. 𝗥𝗲𝘀𝗲𝗿𝘃𝗲 𝘁𝗼 𝗥𝗲𝗱𝘂𝗰𝗲 𝗥𝗮𝘁𝗲: 
Leverage discounted pricing options such as Reserved Instances, Savings plans based on your workloads

𝘉𝘰𝘯𝘶𝘴 𝘛𝘪𝘱:
Spot Instances and lower-cost storage tiers usage isn't a bad idea too

𝟲. 𝗢𝗽𝘁𝗶𝗺𝗶𝘇𝗲 𝗗𝗮𝘁𝗮 𝗧𝗿𝗮𝗻𝘀𝗳𝗲𝗿𝘀:
- Leverage data compression and CDNs to lower bandwidth charges
- Minimize data transfer costs with smart resource placement and intra-region transfers

𝗡𝗼𝘁𝗲:
It's recommended to carefully consider the impact on your specific workloads and seek expert advice as necessary

Need help ? Let's talk !!

🔁 Consider a Repost if this is useful.

=============================
Follow me, tap the (🔔) on my profile Govardhana Miriyala Kannaiah

You'll be notified the moment I post

Let's make your Digital and Cloud transformation journey seamless 🚀

#gopuwrites #FinOps #CloudCostManagement #CloudFinancialManagement #aws #gcp #azure

I once asked this in an AWS cloud engineer interview.

If I assign a /24 CIDR to a VPC, how many usable IPs are there?

The candidate confidently replied: 256 IPs.

Before getting into whether it was the right or wrong answer,

I see many cloud practitioners, especially beginners, find it hard to choose the right subnet size.

I made this visual to help you understand. Check it out !

A /24 provides 256 IPs – right.

But AWS reserves 5 IPs per subnet.

These include: network address, broadcast, router, DNS, and one more for future use.

So, /24 gives you only 251 usable IPs, not 256.

Plan usable IPs, not theoretical ones.

47K+ read my TechOps Examples newsletter: https://lnkd.in/gg3RQsRK

What do we cover:
DevOps, Cloud, Kubernetes, IaC, GitOps, MLOps

🔁 Consider a Repost if this is helpful

🚀 Kubernetes Periodic Table 🐳 👇

120 𝙘𝙤𝙢𝙥𝙤𝙣𝙚𝙣𝙩𝙨 𝙘𝙪𝙧𝙖𝙩𝙚𝙙 𝙡𝙞𝙨𝙩

🔁 Consider a Repost if this is useful.

=============================
Follow me, tap the (🔔) on my profile Govardhana Miriyala Kannaiah

You'll be notified the moment I post

Let's make your Digital and Cloud transformation journey seamless 🚀

#gopuwrites #Kubernetes #DevOps #Cheatsheet

90% of DevOps roadmaps are inaccurate.

Typical roadmaps look like:

Containerization: Docker, Helm, Kubernetes
CI/CD: Jenkins / Github actions / GitLab
Programming: Go / Python
Cloud: AWS / Azure / GCP
Version Control: GIT
IaC: Terraform
CM: Ansible

There is a pre-state to this.

I call it 'DevOps Day 0' Roadmap:

Networking: TCP/IP, DNS, HTTP/S, VPN, Load Balancers, Firewalls, Network Protocols, Subnetting

Database: SQL vs. NoSQL, ACID Properties, Scalability, Data Modeling

Security: Encryption, Authentication, Authorization, OWASP Top 10, Security Policies, Risk Assessment, Compliance Standards (like GDPR, HIPAA).

Storage: Block Storage, Object Storage, File Storage, NAS, SAN, SSD vs. HDD.

DR: Backup and Restore, Pilot Light, Warm Standby, Multi-site, RTO (Recovery Time Objective), RPO (Recovery Point Objective).

Data Replication: Master-Slave, Peer-to-Peer, Synchronous vs. Asynchronous, Data Consistency, Replication Topologies, Log Shipping.

Cache: In-memory Caches (Redis, Memcached), CDN, Cache Invalidation, Write-through vs. Write-back Cache, Cache Hit Ratio.

DevOps is 20% building, 80% optimizing and operating.

Get the 'Day 0' basics right before jumping into tools.

10000+ read my free, bite-sized weekday TechOps examples newsletter (Mon-Fri): https://lnkd.in/gg3RQsRK

What is covered:
DevSecOps, Cloud, Containerization, IaC, GitOps, MLOps

🔁 Consider a Repost if this is helpful.

How I Reduced Docker Image Size from 588 MB to Only 47.7 MB

𝗔 𝘄𝗵𝗼𝗺𝗽𝗶𝗻𝗴 𝟵𝟭.𝟴𝟵 %

To begin with, there is no secret here if you already know about the multi stage builds.

We all know minimizing docker image sizes accelerates container deployment, and for large-scale operations, this can lead to substantial savings in storage space.

1. For a flask app, I picked up 𝗣𝘆𝘁𝗵𝗼𝗻 𝟯.𝟵-𝗮𝗹𝗽𝗶𝗻𝗲 𝘄𝗵𝗶𝗰𝗵 𝗶𝘀 𝗮 𝘄𝗵𝗼𝗺𝗽𝗶𝗻𝗴 𝟵𝟱.𝟮% 𝘀𝗺𝗮𝗹𝗹𝗲𝗿 𝘁𝗵𝗮𝗻 𝗣𝘆𝘁𝗵𝗼𝗻 𝟯.𝟵

This minimal images contain only the essentials, significantly reducing the image size.

2. I minimized layers - every command in a Dockerfile (like RUN, COPY, etc.) generates a separate layer in the final image.

Grouping similar commands together into one step makes sense, which decreases the total number of layers, leading to a smaller overall image size.

𝗜𝗻𝘀𝘁𝗲𝗮𝗱 𝗼𝗳 𝗱𝗼𝗶𝗻𝗴 𝘁𝗵𝗶𝘀:

RUN apk update
RUN apk add --no-cache git
RUN rm -rf /var/cache/apk/RUN apk update
RUN apk add --no-cache git
RUN rm -rf /var/cache/apk/* *

𝗗𝗼 𝘁𝗵𝗶𝘀:

RUN apk update && apk add --no-cache git && rm -rf /var/cache/apk/*

3. Used .𝗱𝗼𝗰𝗸𝗲𝗿𝗶𝗴𝗻𝗼𝗿𝗲 File - Docker transfers all the files from your project directory into the image by default. To avoid including unneeded files, used a .dockerignore file to exclude them.

__pycache__
*.pyc
*.pyo
*.pyd
venv/

4. Multi-Stage Builds - Here all the magic happens !

Dockerfile looks like:

# Stage 1: Build
FROM python:3.9-alpine AS builder

# Install necessary build dependencies
RUN apk add --no-cache build-base \
 && apk add --no-cache gfortran musl-dev lapack-dev

# Set the working directory
WORKDIR /app

# Copy the requirements file and install dependencies
COPY requirements.txt ./
RUN pip install --no-cache-dir -r requirements.txt

# Copy the rest of the application code to the working directory
COPY . .

# Uninstall unnecessary dependencies
RUN pip uninstall -y pandas && apk del build-base gfortran musl-dev lapack-dev

# Stage 2: Production
FROM python:3.9-alpine

# Set the working directory
WORKDIR /app

# Copy only the necessary files from the build stage
COPY --from=builder /app /app

# Expose the port the app will run on
EXPOSE 5000

# Run the Flask app
CMD [“python“, “app. py“]

The new image size is: 𝗢𝗻𝗹𝘆 𝟰𝟳.𝟳 𝗠𝗕

Single stage image size was: 𝟱𝟴𝟴 𝗠𝗕

The application works exactly the same, but it spins up much faster in this version.

That's an whomping -𝟵𝟭.𝟴𝟵 %

👋 PS - I wrote an article with visuals and detailed break down of this use case in my newsletter here: https://lnkd.in/gFzjeCXu

🔁 Repost if this is helpful

Linux → Docker → Networking → Kubernetes

Not the other way around.

Linux → For strong fundamentals

Docker → For containerizing applications

Networking → For connecting everything together

9000+ read my free bite-sized weekday (Mon-Fri) daily TechOps examples newsletter

What do we cover:
DevSecOps, Cloud, Containerization, IaC, GitOps, MLOps

Join us here for FREE: https://lnkd.in/gg3RQsRK

🔁 Consider a Repost if this is helpful

Not a joke, many DevOps Engineers don't fully understand what DevOps truly is.

Learning tools is just a skill.

You would become a DevOps Engineer when you understand: 

How to handle unplanned outages ?
How a change impacts the business ?
What are the dangerous blind spots in the system?
How to collaborate with stakeholders during a crisis?
How can the system run independently in your absence?

I could keep going. 

With just skills: 

You may do well in some teams, sometimes.  
You may do well in many teams, sometimes. 

Only with expertise: 

You would do well in many teams, many times. 

Aim for expertise.

I repeat,

DevOps skills ≠ DevOps expertise

32K+ read my free bite-sized weekday (Mon-Fri) daily TechOps examples newsletter: https://lnkd.in/gg3RQsRK

What do we cover:
DevOps, Cloud, Containerization, IaC, GitOps, MLOps

🔁 Consider a Repost if this is helpful

GitHub is free.
Ansible is free.
Docker is free.
Terraform is free.
Minikube is free.
AWS* is free (to an extent).
Python is free.
ArgoCD is free

You can build a strong DevOps and Cloud career with nothing but a laptop and internet.

No more excuses.

29K+ read my free bite-sized weekday (Mon-Fri) daily TechOps examples newsletter: https://lnkd.in/gg3RQsRK

What do we cover:
DevSecOps, Cloud, Containerization, IaC, GitOps, MLOps

🔁 Consider a Repost if this is helpful.

Linux → Networking → Docker → Kubernetes

Not the other way around.

Linux → For strong fundamentals

Networking → For connecting everything

Docker → For containerizing applications

16K+ read my free bite-sized weekday (Mon-Fri) daily TechOps examples newsletter: https://lnkd.in/gg3RQsRK

What do we cover:
DevSecOps, Cloud, Containerization, IaC, GitOps, MLOps

🔁 Consider a Repost if this is helpful

You're not a Kubernetes Pro until you have…

Watched a pod fail over and over with no idea what’s wrong

Had at least one nightmare about pods in CrashLoopBackOff

Cluster crash in the middle of the night, forcing you to fix it half asleep

I can go on...

44K+ read my free newsletter: https://lnkd.in/gg3RQsRK

What do we cover:
DevOps, Cloud, Kubernetes, IaC, GitOps, MLOps

🔁 Consider a Repost if this is helpful

DevOps and Cloud Engineering is not just about learning tools.

Learn basic working blocks for the long run:

1. How does networking work?
- Keywords: TCP/IP, DNS, HTTP/S, VPN, Load Balancers, Firewalls, Network Protocols, Subnetting

2. How to pick the right database?
- Keywords: SQL vs. NoSQL, ACID Properties, Scalability, Data Modeling

3. Principles of secure system design
- Keywords: Encryption, Authentication, Authorization, OWASP Top 10, Security Policies, Risk Assessment, Compliance Standards (like GDPR, HIPAA).

4. What are the different kinds of storage?
- Keywords: Block Storage, Object Storage, File Storage, NAS, SAN, Cloud Storage (AWS S3, Azure Blob), SSD vs. HDD.

5. What are different kinds of DR (Disaster Recovery) strategies?
- Keywords: Backup and Restore, Pilot Light, Warm Standby, Multi-site, RTO (Recovery Time Objective), RPO (Recovery Point Objective).

6. Common approaches for data replication
- Keywords: Master-Slave, Peer-to-Peer, Synchronous vs. Asynchronous, Data Consistency, Replication Topologies, Log Shipping.

7. When to implement caching mechanisms?
- Keywords: In-memory Caches (Redis, Memcached), CDN, Cache Invalidation, Write-through vs. Write-back Cache, Cache Hit Ratio.

Tools and frameworks will change. Fundamentals remain intact.

Without fundamentals, you are incomplete.

Think about it.

Dear Managers:

DevOps is a tough job.

- Fixing issues fast
- Automating tasks
- Keeping things running smoothly

Then suddenly, they are handling way more than expected.

More work?

That’s fine!

But without enough resources?

❌ Big trouble ❌

That’s when things go wrong.

→ Be clear about priorities.  
→ Be clear on what you really need  
→ Focus on what’s important, not everything.

Your aspirations to move mountains are fine.

But not at the cost of their burnout.

Great work ≠ Infinite energy

Don’t wait for problems to happen:

Support your DevOps team.

They’re problem solvers, not magicians 🛠️

12,000+ tech pros from Apple, Amazon, Nike, Stripe read my free weekday daily TechOps examples newsletter : https://lnkd.in/gg3RQsRK

What do we cover:
DevSecOps, Cloud, Containerization, IaC, GitOps, MLOps

🔁 Consider a Repost if this is helpful.

Don't build CI/CD pipelines that JUST work.

Build ones that are intelligent and efficient.

Just a pipeline running successfully doesn't mean it's the end of it.

It's not just about getting from point A to point B; it's about how efficiently you can make the journey.

I personally would like to see intelligent pipelines leveraging:

Parallel build ⟶ To speed up build time  
Automated rollback ⟶ To recover quickly  
Jenkins as a slave ⟶ To distribute workloads
Containerized builds ⟶ To ensure consistency
Pipeline as code ⟶ To improve maintainability  
Caching mechanisms ⟶ To reduce redundancy  
On-demand agents ⟶ To optimize resource usage 

Remember,

The goal of a CI/CD pipeline is not JUST successful deployments.

But faster, secure, reliable, and maintainable deployments.

Think about it.

Pro level DevOps folks would find a gap in the example. Share it in the comments if you find 😀

🔁 Consider a repost if this is helpful.
--
I'm Govardhana Miriyala Kannaiah and I run a digital and cloud consulting firm, NeuVeu. Trusted by Stanford University, Hearst Corporation and 15+ clients in 12+ countries over the last year.

If you are a business looking for help, you can schedule a free business consultation call with me (calendly link in the bio)