Carlos Nunez

Writing greener code doesn't start with micro-optimizations. It starts with staying up to date. In the latest Environment Variables episode, Sara Bergman (Microsoft) and host Anne Currie discuss the practicalities of greener computing, from keeping your platforms up to date to transitioning .NET services to ARM64 architectures. Learn how to get the cost and performance gains from architecture migrations and why the ability to make changes matters more than perfect code. Tune in to the full episode ➡️ https://lnkd.in/dJyzDVcr

52

9 Comments

Terraform Automation Framework for Enterprise AWS Platforms Infrastructure as Code is essential for operating large-scale AWS environments consistently. In enterprise environments supporting hundreds or thousands of AWS accounts, Terraform helps enforce standardization, repeatability, and governance. Some key practices that have worked well in our environment: • Modular Terraform architecture for reusable infrastructure components • Standardized Terraform modules for networking, IAM, and shared services • GitHub Actions CI/CD pipelines for automated infrastructure deployment • Policy guardrails integrated into infrastructure provisioning workflows • Automated compliance checks and drift detection By implementing these patterns, infrastructure provisioning becomes predictable, secure, and scalable across large AWS environments. Terraform combined with CI/CD enables organizations to move toward fully automated cloud platforms. Curious to learn how other teams structure Terraform frameworks at scale. #AWS #CloudArchitecture #PlatformEngineering #DevSecOps #FinOps

7

2 Comments

The biggest gain after rolling out observability in our engineering practice? Visibility into how much of our architecture behaved differently in a production setting than we expected. Our platform ran primarily on AWS, with a small portion still hosted in a traditional data center, a mix of .NET services and React/.Net applications, each with its own logging and alerting setup. Often by the time we knew something was wrong, users already had a bad experience. I led the effort to design and deploy a unified observability framework across the stack using New Relic and working closely with engineering, DevOps, and infrastructure teams. We instrumented key AWS services, APIs, front end applications, and database calls, built dashboards aligned to DORA metrics and business KPIs, integrated alerts into Teams and CI/CD, and set up tagging and baselines for faster SRE response. We also leveraged service maps, distributed tracing, and synthetic transactions to visualize dependencies and detect issues before they affected production. 𝗧𝗵𝗲 𝗿𝗲𝘀𝘂𝗹𝘁𝘀:   • We could trace issues from user session → API → database in seconds.   • Engineers stopped scrambling to dig for the root of the issue, we had data.   • Shifted teams from reactive firefighting to proactive optimization   • Improved customer experience scores due to fewer latency spikes and smoother user journeys. 𝗛𝗼𝘄 𝘄𝗲 𝗺𝗮𝗱𝗲 𝘁𝗵𝗶𝘀 𝘄𝗼𝗿𝗸:   • Establishing telemetry as a facet of our maturity model, evolving from simple monitoring to end-to-end visibility.   • Using distributed tracing to connect transactions across services (internal and 3rd party) and identify hidden bottlenecks.   • Visualizing dependencies through service maps to align technical metrics with business outcomes.   • Embedding observability into DevOps workflows to reduce MTTR, improve collaboration, and optimize costs across AWS and hybrid environments. Observability via telemetry is not only about catching errors faster; it is about creating data driven feedback loops that make engineering more predictable and business conversations more grounded in reality. How have you approached observability in your own systems, any lessons learned along the way?

2

1 Comment

🚀 Exciting news! I’ll be speaking at AWS re:Invent 2025 in Las Vegas 📅 December 1–5, 2025 📍 Las Vegas, NV 🎤 Session Code: MAM310 ✨ AWS Migration planning: From discovery to delivery Cloud migration at enterprise scale is never just “lift and shift.” Join me for this chalk & talk session where we’ll dive deep into the end-to-end journey of migration planning: 🔍 Running effective discovery & environment assessments 🧩 Building robust migration strategies that balance tech and business needs ⚡ Accelerating analysis with AWS Transform 📊 Turning plans into actionable execution models 🌎 Learning from thousands of real-world AWS transformations If you’re heading to re:Invent and want to sharpen your migration strategy—or avoid common pitfalls—this session is for you. 💡 Let’s connect, share experiences, and shape what successful cloud journeys look like in 2026 and beyond. ✈️☁️ #AWS #reInvent #CloudMigration #Modernization #MultiCloud

113

5 Comments

Terraform vs. Lime Boost: What's the right tool for your cloud infrastructure? Terraform is a powerful, flexible, and battle-tested tool for Infrastructure as Code (IaC). It gives you granular control over every aspect of your cloud resources, making it the top choice for complex, bespoke environments and expert DevOps teams. But what if you don't need that level of complexity? What if you just need to get things done, fast? That's where Lime Boost comes in. We're built for developers and small teams who want the power of AWS without the IaC and AWS learning curve. Lime Boost automates routine tasks like deployment and resource cleanup, so you can focus on building your product, not managing your infrastructure. #Terraform #AWS #CloudComputing #SaaS #Startup

5

AWS | DynamoDB us-east-1 Outage ~ Postmortem AWS recently published a detailed postmortem on the October 19-20, 2025 DynamoDB outage in us-east-1. As someone who follows distributed systems reliability, I found this analysis incredibly valuable. The Root Cause: A race condition in DynamoDB's DNS management automation caused an older DNS plan to overwrite a newer one. When the cleanup process deleted that older plan, it removed ALL IP addresses for the regional endpoint—making DynamoDB completely unreachable. The Cascade: DynamoDB down → EC2 launches failed (lease management depends on DynamoDB) EC2 issues → Network state propagation backlog Network delays → NLB health check failures Result: Lambda, ECS, EKS, Fargate, Connect, STS, and dozens of other services impacted Key Lessons: - Race conditions in distributed systems are incredibly hard to catch - Even the world's most sophisticated systems will eventually fail. The true measure of engineering excellence isn't preventing every failure—it's how quickly and effectively you can navigate to the root cause. AWS's Response: They've disabled the automation globally, are fixing the race condition, adding protections against incorrect DNS plans, and building better scale testing for recovery workflows. This is a textbook example of how even the most sophisticated systems can fail in unexpected ways. Kudos to AWS for the transparent postmortem. Full details: https://lnkd.in/gU9xcapN #AWS #DistributedSystems #PostMor

13

3 Comments

Ever wondered how to build a truly data-driven autoscaling strategy for Kubernetes? 🚀 Our latest video on YouTube dives deep into Kubernetes autoscaling, with plenty of hands-on tips and tools for scaling smarter, not harder. Here's what you'll learn: 📈 The difference between HPA & VPA (and when to use each) 💣 How to avoid misleading default metrics in your scaling setup 📊 Using external metrics like traffic volume for more precise scaling 🌐 Correlating resource usage with application health via OpenTelemetry and Dynatrace 🌟 Automating scaling decisions with tools like DQL, EdgeConnect, and GitOps workflows We’ve packed this episode with practical insights on tools like KEDA, KEPN, and the Dynatrace Operator, so you can simplify scaling processes and improve application performance. Head over to our YouTube channel to watch the full video. Don’t miss out! #kubernetes #autoscaling #opentelemetry

5

🚀 Day 9 of #100DaysOfDevOps – MariaDB Troubleshooting 🐬💥 Today’s challenge was a critical one — the Nautilus application in Stratos DC couldn’t connect to the database. After investigation, the issue turned out to be that the MariaDB service was down on the database server. 🔍 I approached it methodically: SSH’d into the DB server and confirmed that mariadb was inactive. Checked the logs via: journalctl -xeu mariadb Identified that the root cause was either: Missing or misconfigured data directory (/var/lib/mysql) Or permission issues with /run/mariadb 🔧 Solution: Recreated the missing mysql data directory Set correct permissions: chown mysql:mysql /run/mariadb systemctl restart mariadb Check the result with: mysql -u root ✅ Result: MariaDB is back online, and the application can connect again smoothly. 📘 This was a great hands-on exercise in real-world database recovery and service troubleshooting — skills every DevOps engineer must have! #DevOps #MariaDB #Linux #SystemAdministration #Troubleshooting #KodeKloud #100DaysOfDevOps #LearningByDoing #DatabaseRecovery

16

7 Comments

Most architecture diagrams end up as static files, leaving #DevOps to manually rebuild them in #Terraform, often taking 8–15 hours per environment. With Codespell’s Design to Code, that work happens in minutes. Upload a diagram, get clean, modular Terraform, the hours you save here compound across every release. Read full blog here https://lnkd.in/gUDugcNR #InfrastructureAsCode #DevOpsAutomation #CloudInfrastructure #IaC #DevOpsTools #InfraAsCode #CloudEngineering #CI_CD #Codespell

14

Just wrote a tutorial using my content MCP server (I call it Ally), which generates an excellent article on GreptimeDB’s TQL + CTE queries for typical Kubernetes monitoring. The best part? It automatically validates all SQL statements itself through the GreptimeDB MCP server. #PowerOfAI #MCP #GreptimeDB

13

Infrastructure challenges only get harder after deployment. That's where Terraform actions bring real value. By standardizing more of your Day 2 workflows within the Terraform pipeline, you can confidently manage infrastructure over time without the manual overhead. A perfect example? Pairing Ansible with Terraform actions for seamless, automated infrastructure and configuration management. Ready to transform how you operate? https://bit.ly/3WKwCk6

54

ClickHouse doesn't have built-in schema migration support. In 2026. For a database used in production by thousands of companies. https://lnkd.in/gCU-4Fry That's not a niche edge case. Schema changes are one of the most common operations in any production database. And in ClickHouse, what looks like a simple column addition can trigger a full table rewrite when you're streaming hundreds of megabytes per second. Tinybird's own migration story is telling. Their first approach to handling materialized view changes was, in their words, "bulletproof but brutally inefficient." Touch just one materialized view and you'd migrate terabytes of upstream data. For a 14TB Kafka table, that meant days-long deployments. Every database has tradeoffs. ClickHouse is genuinely fast for narrow analytical queries. But if your workload evolves (and it will), the operational cost of that evolution is worth understanding before you're 14TB deep. What's the most painful schema migration you've dealt with?

7

Kubernetes is a means, not the end goal. It’s easy to get lost in the complexity of clusters and service meshes, but we must remember: the goal isn't "to run Kubernetes," it's to deliver value through our applications. Infrastructure should exist to empower software, not to complicate it. While an application must be architected to leverage cloud-native capabilities, the technology should never become a bottleneck for shipping features. When the "how" starts to slow down the "why," we’ve lost our way. As engineers, our ultimate priority isn't managing complexity. it's writing the code that solves actual problems for our customers. Infrastructure is the foundation, but the application is where the impact happens. Where do you draw the line between technical excellence and over-engineering? I know we all like shiny new tooling.. Do you know that we have a De Nederlandse Kubernetes Podcast? Tell us in our podcast what your new tooling is. #Kubernetes #CloudNative #PlatformEngineering #SoftwareDevelopment #podcast

10

4 Comments

AI can generate a Terraform module, a Bicep template, or a full Landing Zone scaffold in seconds. It looks right. It deploys. But that's where the trouble begins. Misconfigured resources quietly drain budgets. Compliance gaps slip past review because the code reads as authoritative. And once AI agents with MCP servers start applying changes on their own, the blast radius is no longer a human typo. Vladimir Stefanovic has spent 20+ years building cloud systems. Microsoft Azure MVP. The kind of architect who starts from business constraints, not the latest model. He's seen the failure modes up close. At Doghouse Unleashed #5, he's covering the questions teams skip when adopting AI-assisted IaC quickly. The ones that don't show up in demos, but decide what happens in production. If you work with Terraform, Bicep, or Azure at scale, this one is for you. June 11. Amsterdam. Free to attend. Sign up: https://lnkd.in/eTPqU7Gf

12

Containment strategy in tech: 1. Find the issue. 2. Contain the issue. 3. Realize the issue contained you instead. #TechHumor #DevOps #Containment

12