mkdev
414 subscribers
861 photos
179 videos
1.07K links
Download Telegram
#mkdevWeeklyHighlight
What you would usually use a front end framework for can now be achieved with CSS and occasional sprinkles of vanilla JavaScript. How is this related to cloud infrastructure and devops? Well, lack of framework means simpler builds, faster delivery and less moving pieces, dependency management and maintenance: https://www.jonoalderson.com/conjecture/its-time-for-modern-css-to-kill-the-spa/
This media is not supported in your browser
VIEW IN TELEGRAM
Episode 62 of DevOps Accents with Mark Fussell from Diagrid & Dapr available now! The place of sidecar architecture now and in the future, and many other topics discussed: https://mkdev.me/posts/kubernetes-isn-t-enough-with-mark-fussell-from-diagrid-dapr-62
In the 76th mkdev dispatch Pablo makes a list of themes and topics to look out for in the coming months. The fall is always big for our industry! Don't miss your chance to make a splash. Subscribe to be the first to get the next dispatch: https://mkdev.me/posts/summer-is-over-76
The third article of this series by Paul Larsen warns product managers about the major cybersecurity risks of GenAI—like data leaks, prompt jailbreaks, and injection attacks—and offers practical steps to keep AI use productive without endangering company data or operations: https://mkdev.me/posts/genai-security-risks-for-product-managers-dd73bdc2-4f2e-4227-93b3-375da081d906
#mkdevWeeklyHighlight
What does it take to crawl 1 billion pages in 2025? On AWS, following the architecture the author came up with: 25 hours; $500 and 12 EC2 instances. Check out the full article to see how: https://andrewkchan.dev/posts/crawler.html
Media is too big
VIEW IN TELEGRAM
Prompt engineering is the new English.
Words = interface to AI.

Now language skills matter more than ever when working with models and agents. In this DevOps Accents episode with Suzanne EL-Moursi we unpack Data Science 2.0: https://mkdev.me/posts/data-science-2-0-with-suzanne-el-moursi-from-brighthive-63
#DataScience #AI #DevOps #podcast #DevOpsAccents
Docker made containers really popular. It was the OCI standards that made them portable. But running a single container is just the start. When you're working on a large scale, you also need orchestration, networking, storage — and it's best to have standards for all of these.

In this wrap-up of the Dockerless series we cover:
🔹 CRI – swap container managers in Kubernetes without having to change Kubernetes itself.
🔹 CNI is the standard for container networking – just plug and play!
🔹 CSI – making storage and orchestration platforms separate
🔹 Kata Containers are lightweight virtual machines that act like containers.
👉 The bigger picture: containers, VMs, networking, and storage are all coming together into one standardised, open ecosystem for building infrastructure.

https://mkdev.me/posts/dockerless-container-standards-landscape

#Dockerless #Containers #Kubernetes #CloudNative #DevOps #Infrastructure #CNI #CSI #CRI #KataContainers #ContainerOrchestration #CloudComputing
#mkdevWeeklyHighlight
This is seriously cool stuff that will simplify certificate management a lot and remove at least one component from your K8s clusters: https://blog.nginx.org/blog/native-support-for-acme-protocol
This media is not supported in your browser
VIEW IN TELEGRAM
How to assess your error budget for core cognitive abilities? In episode 64 of DevOps Accents, Paul Larsen, our Head of Data & AI, is back to discuss with Leo recent studies from Microsoft and MIT about ChatGPT and critical thinking: https://mkdev.me/posts/chat-gpt-is-making-you-dumber-not-a-clickbait-there-is-a-research-for-that-64
“Human in the loop” sounds smart—until it’s just a queue of people approving every answer. Paul Larsen reframes operational maturity: if your error budget allows autonomy, run it like a production ML system—with telemetry, thresholds, and audits, not endless approvals. Hear the full conversation on DevOps Accents—get it wherever you get your podcast! https://www.youtube.com/watch?v=PKaqKgm2jVY
#mkdevWeeklyHighlight
This is great, we don’t need to make everyone install extra tooling required to open up console sessions to our containers: https://aws.amazon.com/about-aws/whats-new/2025/09/ecs-exec-aws-management-console/
This media is not supported in your browser
VIEW IN TELEGRAM
Why are the reasoning models currently the default? They are too slow! More on how our brain interacts with Chat GPT and others in episode 64 of DevOps Accents with Paul Larsen: https://mkdev.me/posts/chat-gpt-is-making-you-dumber-not-a-clickbait-there-is-a-research-for-that-64
From Public Cloud to IaC, from Data Engineering to AI Literacy, from Pipelines to Platform Engineering: check out the full list of mkdev Consultancies and schedule a call with us: https://mkdev.me/b/consulting

Innovate faster with zero compromises
#mkdevWeeklyHighlight
As always, major Postgres releases bring a load of new features and improvements. Tudor Golubenco from Xata does a great job of giving us an overview together with practical examples: https://xata.io/blog/going-down-the-rabbit-hole-of-postgres-18-features
In the 79th mkdev dispatch, Pablo talks about the last days of “being a developer” being a thing. Also inside: cross-account and cross-region log centralization in CloudWatch, Terraform Optimization Guide, CloudFlare Radar AI Insights and more! https://mkdev.me/posts/when-developer-is-no-longer-a-profession-79
#mkdevWeeklyHighlight
This is smart: use S3 to transfer big amounts of data between A and B, due to the way S3 is priced. Definitely useful for big migrations that would otherwise cost a lot of data egress fees: https://www.bitsand.cloud/posts/slashing-data-transfer-costs/
We’ve seen every kind of infra and team setup—many we’ve designed or tuned. There’s always room to level up. We know what “great” looks like and can get you there. Book a call: https://mkdev.me/b/audits