Vishnu Bharathi

A friend of mine smashed every other candidate in his DevOps application by doing what 99% of people don’t: ❌ No fancy resume ❌ No buzzword salad ✅ Just real, working infrastructure Here’s what he submitted on GitHub: 1. A Terraform module to provision AWS infra (VPC, EC2, S3, IAM) 2. A CI/CD pipeline (GitHub Actions) to deploy a containerised app to ECS 3. Monitoring via Prometheus + Grafana (with alerting rules) 4. Infrastructure tests using Terratest 5. All automated with a Makefile + clear folder structure 6. He also hosted the app on a real domain, as an example: app.<company>.<his-name>.com To top it off, he wrote a 2-page PDF explaining how to run, test, and deploy everything locally or on AWS with minimal setup. Guess what happened? 📞 The company called him within 24 hours. 💬 They said it was the most impressive application they’d ever received. No nonsense. No job board roulette. Just proof of work. Everyone’s out here mass applying with templated resumes. Meanwhile, this guy built the job he wanted. ⸻ 💥 Proof > Promise 💥 GitHub > CV 💥 Shipping > Talking ⸻ Thinking of breaking into DevOps? This is how you stand out. Not by adding “CI/CD” to your resume, but by showing you can build it. #systemdesign #devops #interviewtips

243

19 Comments

Booking(dot)com increased >250,000,000 Reviews backend hardware by 50% Handling ~10,000 req / sec, maintaining p99 of < 50ms Background Review data is partitioned across several shards, with property ID as partitioning Key Entire system is replicated across several availability zones Problem Changing the number of shards means the keys will be redistributed across the remaining shards. if the routing layer tries to resolve the shard ID (of a given accommodation) during the ongoing resharding, it would not be able to resolve the location Solution - They build a service using "Consistent Hash Algorithm by John Lamping, Eric Veach" that ensures randomly distributed data and allows for scaling in both directions by moving bare minimum data around shards - The property that makes it possible is Monotonicity that means data can only be transferred from old shards to new shards during an ongoing resharding. (no unnecessary rearrangement) - Until the new shards have fully loaded the reassigned keys, nodes of our routing layer are only aware of the old scheme and route traffic only to the old shards - After new shards have loaded reassigned keys, some of the coordinators were made aware of the new sharding scheme - After all traffic is routed to the new scheme, old shards are updated to stop loading keys that were reassigned to new shards. - The algorithm is fast (nanoseconds) and has no additional storage requirements 🤔 over to you, how they predicted future hardware requirements Let me know in the comments! 👇 --- You've got System Design skills but showcasing them in 60 minutes interview is a completely different challenge for e.g. where to start? How much time to spend on HLD vs LLD etc Let me share a FREE 7-day email course that provides a step-by-step framework for acing your 60-minute System Design interview 📣 Join 2200+ subscribers and sign up for the course to learn time allocation to demonstrate full range of design skills (link in comments 👇)

101

14 Comments

Namaste 🙏 . Swagat hai aapka - The SRE Times mein! Aaj ki taaza khabar jo apki next interview crack karne meain madad karegi. Breaking News: Amazon Web Services (AWS) just announced GA of the Secrets Store CSI Driver provider as an Amazon EKS add-on. 🤗 What does this mean? This new integration allows customers to retrieve secrets from AWS Secrets Manager and parameters from AWS Systems Manager Parameter Store and mount them as files on their Kubernetes clusters running on Amazon Elastic Kubernetes Service (Amazon EKS). The add-on installs and manages the AWS provider for the Secrets Store CSI Driver. Follow these easy Steps: 1. Install AWS Secrets Store CSI Driver provider add-on 2. Create IAM role with trust policy for pods.eks.amazonaws.com 3. Attach AWSSecretsManagerClientReadOnlyAccess policy to IAM role 4. Install EKS Pod Identity Agent add-on 5. Create Pod Identity Association (links service account to IAM role) 6. Create and deploy SecretProviderClass YAML file 7. Deploy pod with CSI volume configuration 8. Verify secret is mounted at /mnt/secrets-store path Key Benefits: ✅ No hard-coded credentials ✅ Automated lifecycle management ✅ In-memory secret storage (tmpfs) ✅ Native EKS integration

157

7 Comments

Redis Eviction Policies a must know for system design : When using Redis as a cache, memory is finite. So the real question becomes what happens when Redis runs out of memory? That’s where Eviction Policies come in. Redis eviction policies define which keys are removed when memory is full and a new write arrives. * Common Eviction Policies -> LRU (Least Recently Used) – evicts keys not accessed recently -> LFU (Least Frequently Used) – evicts keys used least often -> TTL (Time To Live) based eviction – evicts keys with lowest remaining TTL -> Random eviction – evicts random keys -> No eviction – writes fail when memory is full * allkeys vs volatile allkeys → any key can be evicted (pure caching use cases) volatile → only keys with TTL are evicted  Production best practice Most modern systems prefer: allkeys-lfu because it protects frequently accessed (hot) data and avoids cache pollution.

70

2 Comments

This team spent 3 weeks debugging a mysterious DNS issue in their Kubernetes staging environment. The symptoms were maddening: → Metrics looked completely normal → Pods were running smoothly → Logs showed zero errors → But certain services intermittently failed to resolve names 3 weeks of investigation. Multiple senior engineers pulled in. Deep dives into networking configurations. Packet captures analyzed. Service mesh debugging sessions. Theories about load balancer issues. The culprit? kube-dns had a default CPU limit set too low. Under moderate load, it would throttle without throwing any alerts or writing error logs. Just... silent failures. This perfectly captures the hidden complexity trap of modern infrastructure and why our current tooling is not scaling in the age of AI. Defaults that seem reasonable but fail under real-world conditions. Critical services that degrade gracefully until they don't. No observability into the actual problem. Debugging and RCA takes hours. Your DNS resolver was quietly dying, but all your monitoring said everything was fine. 3 weeks of engineering time lost to a CPU limit that nobody questioned because it was the "recommended default." Modern infrastructure hides its failures behind layers of abstraction. The error messages don't exist. The metrics look healthy. The logs are clean. But your applications randomly break. How many weeks have you spent debugging "impossible" problems?

74

29 Comments

🚗 Ever wondered how #Uber handles millions of real-time requests across the globe, without crashing? Uber System Design: Scalable, Modular & Real-time Uber’s architecture is built to support real-time operations across the globe, using modern cloud-native tools and layered system components. 1. IaC (Infrastructure as Code): Uber uses Terraform to automate infrastructure provisioning, ensuring consistency and speed across cloud environments. 2. Containerization & orchestration: Docker and Containerd manage containerized services, while Kubernetes handles scaling, availability, and service orchestration. 3. Databases: A diverse stack including Cassandra, MySQL, Redis, Hadoop, and Docstore supports real-time, analytical, and document-based data needs. 4. Cloud storage: Uber leverages Amazon S3, Azure Storage, and Google Cloud to store and access data across multiple cloud providers efficiently. 5. Microservices & caching: Built on microservices, the platform uses Redis-backed query and storage engines to reduce latency and ensure fast data access. 6. Big data platform: Kafka ingests data into systems like Hive, Spark, and Presto for large-scale processing and business analytics. 7. Observability & monitoring: Grafana, Jaeger, Prometheus, and other tools offer full visibility into performance, logs, and service health. [Explore more in the post] In summary: Uber’s system design reflects a robust, multi-cloud, real-time infrastructure, optimized for scale, speed, and seamless user experience across millions of rides and data interactions. 🔁 Save this for reference and share with someone preparing for system design interviews! 👉 DM me for any mentorship program to successfully land your dream tech role. Everyone Who Codes (EWC) has a simple mission: helping engineers get hired! 💼 Follow Lakshmi Marikumar for tech career & job search insights. Subscribe to our free newsletter on Substack: https://lnkd.in/gN83YA4H #systemdesign #softwareengineering #techjobs #interviews #backend #softwareengineer #softwaredeveloper #MAANG #2ndtier #hiring #jobsearch #uber  

194

21 Comments

Stuck keeping #Spring apps alive on #EC2 (or herding #Kubernetes) just to ship a few endpoints? ⚙️ What if #Java on #AWS #Lambda could meet your latency targets and cut the ops tax too? ⏱️💸 Good news! Serverless can be your escape hatch... Yes, Java runs great on AWS Lambda. You can hit realistic p95s, keep costs sensible, and still have a genuinely pleasant developer experience. ☕️ In this fresh episode of AWS Bites, I sit down with Mark Sailes (ex-AWS, Java-on-Lambda expert). We get practical: when Java is the right call for Lambda and when it isn’t, how SnapStart and provisioned concurrency actually help, which frameworks behave at startup (vanilla, #Micronaut, #Quarkus, Spring Boot), and how to test without drowning in brittle mocks (#Testcontainers, #LocalStack). 🧪🔧 If your stack feels like a cage, this episode points to the exits. 🧭🚪 🎧 Watch/listen: https://awsbites.com/148 Time to migrate your Java services to Lambda, let's get to work! 🚀 PS: Hope you enjoy the slightly hallucinated cover... no substances involved, unless you count the AI’s triple espresso. ☕️🤖 #Serverless #AWSBites #SnapStart #ProvisionedConcurrency #GraalVM #CloudEngineering

39

7 Comments

"If you're struggling with Linux kernel internals, device drivers, or handling interrupts — you're not alone. These topics are deep and complex. But every bug debugged, every dmesg printed, and every IRQ handled brings clarity. Just remember — the struggle means you're learning, and your understanding is getting sharper with every step. Keep pushing. You're getting better." #ROS2 #Robotics #DDS #TechnicalCommunication #SoftwareEngineering #DistributedSystems #TechEducation #Engineering #Innovation #TechLeadership #RoboticsDevelopment #Middleware #TechAnalogy #SystemsEngineering #TechTraining #embedded #embeddedsystems #embeddedsoftware #linuxkernel #embeddedlinux #embeddedengineer #linux

45

2 Comments

النهاردة بنسميها 𝐈𝐚𝐂. زمان كنا بنسميها. "𝐉𝐮𝐬𝐭 𝐚𝐧𝐨𝐭𝐡𝐞𝐫 𝐝𝐚𝐲 𝐰𝐢𝐭𝐡 𝐁𝐚𝐬𝐡." Before Terraform, there was a black screen and a blinking cursor. Everyone is an "Infrastructure as Code" expert now, writing elegant HCL and running complex Ansible playbooks. We celebrate the tools. But we forget where the philosophy came from. IaC wasn't born in a fancy UI. It was born in the trenches, with Linux. Before the resource "aws_instance"... there was a Bash script that provisioned a VM from scratch. Before a schedule block in a YAML file... there was a raw #crontab -e entry that never failed. Before ansible-playbook... there was a chain of #ssh commands and #sed to configure a fleet of servers. The tools gave us syntax. Linux gave us the mindset. It taught us that infrastructure should be treated like code: versioned, repeatable, and automated. It showed us the power of a single command line to build an entire world. Today's tools are powerful abstractions. But the engineer who understands the underlying shell scripts, cron jobs, and init systems doesn't just use IaC, they understand it. The tools change. The foundation doesn't. قبل Terraform، قبل Ansible، قبل Puppet وChef… كان فيه Linux. بسيط، مباشر، وبيعلمنا إن التحكم في البنية يبدأ من سطر كود واحد. كل اللي بنسميه النهارده Infrastructure as Code (IaC)، أساسه اتبنى زمان في Linux: الـ init scripts اللي بتشغّل السيرفر من أول ثانية. الـ cron jobs اللي بتنظم التكرار والاستمرارية. الـ Bash scripts اللي كانت بتبني أنظمة كاملة من لا شيء. Linux علّمنا إن البنية التحتية مش مجرد hardware، لكن code ممكن نكرره، نعدله، ونبني عليه. النهارده عندنا أدوات ضخمة زي Terraform وAnsible، لكن الجذور؟ كلها بدأت مع Linux. الدرس الحقيقي: لو فاهم Linux، فاهم IaC… حتى قبل ما الكلمة نفسها تتولد. 🐧 #Linux #IaC #InfrastructureAsCode #Automation #DevOps #PlatformEngineering #SRE #ShellScripting #Ansible #Terraform #Puppet #Chef #CloudComputing #SysAdmin #Kernel #OpenSource #RedHat #Fedora #Containers #CI_CD #CloudNative #Kubernetes #DevSecOps #ZeroTrust #CyberSecurity #LinuxForLife #TechLeadership #Ops #Infrastructure #Observability #PlatformOps #CloudEngineers #SystemsEngineering #Resilience #Innovation #LinuxCommunity #RedHatAccelerators #EngineeringExcellence

63

8 Comments

Don't set unreal expectations for your first Cloud-DevOps job: • You won’t be asked to build a CI/CD pipeline on day 1. • You won’t be updating the YAMLs on day 2. • You won’t be modernising the entire infrastructure with Terraform. • You won’t be automating anything and everything you see. • You won’t be asked to create custom dashboards. • You won’t be asked to replace Jenkins with GitLab. • You won't be handling a full-scale data migration. • You won’t be asked to spend your weekend troubleshooting that late Friday push. Set the expectations low, every company has their unique definition of Cloud-DevOps.

430

21 Comments