SQL Indexing for Real-World Performance: What Every DevOps Engineer Should Know

As DevOps engineers, we often focus on CI/CD, automation, and infrastructure — but database performance can become a hidden bottleneck in production.

I recently made a beginner-friendly breakdown of SQL indexing that keeps it simple, visual, and practical:

Heap tables – what happens when no clustered index exists

Clustered indexes – how data is physically ordered and retrieved

Non-clustered indexes – when to use them and how they reference the table

Stored Procedure Lookups – real performance examples that show why indexing matters in production


👉 The goal: make indexing easy to understand for people who don’t live inside SQL every day, but still need to keep systems running fast and reliable.

Video link here: https://youtu.be/cDiCp64V-uQ?si=qCKHn0hyGd_ID5MM

Would love to hear how you approach database optimization in your DevOps workflow (monitoring, tuning, automation, etc.)

https://redd.it/1npnqut
@r_devops

Struggling with API misconfigurations in prod

Our team keeps running into API problems once code hits production; things like:

* Incorrect auth settings
* Debug endpoints left open
* Tokens that don’t get updated

A lot of it seems to come from drift between what’s in code and what’s actually deployed.

We’ve tried linting rules and CI/CD checks, but some issues still slip through.

For those of you managing complex stacks, what approaches have helped you catch or prevent API misconfigurations without slowing down your release cycle?

https://redd.it/1npsm3j
@r_devops

Dockerhub is down

TL;DR:
Docker Hub is partially down (mainly auth + registry + web). They know the issue and are working on it. Just hang tight, avoid huge CI bursts, and retry periodically.

What’s happening:

The trouble affects Docker Hub Registry, authentication, and web services.

Their status page says they’ve identified the cause and are working on a fix. 

What you can try now:

Retry operations (pulls/login) periodically- sometimes the errors are intermittent.

Use image digests instead of tags (some users report those are still working).

If you have a mirror or pull-through cache (e.g. via a private registry), consider using it.

Monitor Docker’s status page closely for updates.

P.S If you found this update helpful, consider checking out The Dev Dex.

https://redd.it/1nptobe
@r_devops

Should backend-to-database connections use SSL if proxy already has SSL?

If my backend is running behind a reverse proxy (e.g., Traefik/Nginx) that already has SSL/TLS enabled for client traffic, do I still need to enable SSL/TLS on the database connection between the backend and the database server considering when in Docker-compose or K8s the database is running on internal network therefore not exposed to the outside traffic?

https://redd.it/1nq0ixi
@r_devops

Uk salary expectations

I'm currently looking to change jobs due to an impending return to office mandate. I've been proactively applying for roles for around 3 months and am struggling to find anything. Are my salary expectations too high?

I'm currently on ~£65k with 2 yrs DevOps, 2 yrs Platform Engineering and 15 yrs in infra roles prior to that. Ideally looking for a remote role on at least a matching salary. The main thing I want rn is stability. Feedback from the one interview I've had so far is that there were some knowledge "gaps" based on my salary expectations. Have rates dropped over the last 2 years or do I just need to brush up?

https://redd.it/1nq0e77
@r_devops

Deployed MERN app on AWS EC2 – Frontend works, but backend not accessible externally

Hi everyone,

I’m learning AWS by deploying a **MERN full-stack project** on an **EC2 Linux instance**, but I’m stuck with the backend. Here’s what I’ve done so far:

1. Launched an **AWS EC2 instance** (Linux) and connected via SSH.
2. Installed **Node.js** (same version as local).
3. Cloned both frontend and backend repos.
4. **Frontend setup:**
* `npm install` → `npm run build`
* Installed **Nginx**, enabled service
* Copied build files to `/var/www/html`
* Opened inbound rules for ports **80, 443, 7777**
* Frontend works fine on public IP
5. **Backend setup:**
* `npm install` → `npm start`
* Works fine with `curl` [`http://localhost:7777/`](http://localhost:7777/) and `curl` [`http://13.60.42.60:7777/`](http://13.60.42.60:7777/) inside EC2
* But when I try [`http://13.60.42.60:7777/`](http://13.60.42.60:7777/) in my browser (local machine), it doesn’t load
* Tried running with **PM2** → still the same issue

# What I expected

My backend should be reachable at [`http://13.60.42.60:7777/`](http://13.60.42.60:7777/) from my local machine.

# What actually happens

* Works locally inside EC2 with `curl`
* Not accessible externally from browser

I’ve repeated this process 3 times with the same result.

Does anyone know what I might be missing? Could it be related to binding `localhost` vs [`0.0.0.0`](http://0.0.0.0), security groups, or something else?

Thanks in advance! 🙏

https://redd.it/1nq125z
@r_devops

Free Course Complete GitHub Actions Course — From Beginner to Pro!

Hi folks! —

I just released the latest course in my DevOps Beginner to Pro series, this one focused on GitHub Actions!

Video: https://youtu.be/Xwpi0ITkL3U
Companion Repo: https://github.com/sidpalas/devops-directive-github-actions-course/tree/main

The course covers:
- History and motivation for Continuous Integration
- Why GitHub Actions?
- Core platform features
- Advanced platform features
- Consuming GitHub Actions Marketplace actions
- Authoring first-party actions
- Common automation workflows
- Improving the developer experience
- Best practices for using GitHub Actions
- An end-to-end capstone project

Check it out and let me know what you think!

https://redd.it/1nq86lx
@r_devops

Built a EC2 & VM price comparator to save my own sanity

I work as a cloud engineer for a big bank firm in Europe, my job basically consists on conducting proof of concepts for any new tools that we have to implement in our infrastructure so I spent all of my time deploying EKS/AKS clusters and EC2/VMs instances here an there.

I basically got tired of juggling to find the cheapest but still capable EC2 type in the CLI for each test while keeping performance decent. So I built a small site that lets me quickly compare EC2 instance families and prices side-by-side. I did not expect to make it public to be honest, but I thought it could help a fellow devops colleage struggling like I was at the beginning.

It’s minimal—no logo, no cookie banner—, let me know if you guys want any new functionality, I will try to implement it asap when I have some free time. There is also an AMI and VM image search tool. Reserved prices and savings plan are next on the roadmap, let me know what you think. cloudpylon.com


ps: It is deployed on a 3 dollar hetzner server so it might feel a bit slow at times :)

https://redd.it/1nq8tk5
@r_devops

European Pulic Clouds

Hey everyone! Is anyone working with a european public cloud at your company already? My company is currently considering StackIT and Telecom Cloud, bith are German. What are your experiences with the respective european cloud providers so far in the corporate context?

Edit: public instead of pulic

https://redd.it/1nqah5j
@r_devops

Deploy to production with just docker compose up

Hey,

Working on lots of small projects at a startup, I kept running into the same issue: deploying to production is either overkill (Kubernetes) or a hassle (managing your own VPS/EC2).

All I wanted was: if it runs locally with Docker Compose, it should run in production the same way. No new CLIs, no servers to babysit.

So I built a service where you can literally do:

$ docker compose up -d

… and your stack is live in the cloud.

Would love feedback from the community, am I the only one to have this problem?

https://wip.cx

https://redd.it/1nqc29w
@r_devops

Immutable Infrastructure DevOps: Why You Should Replace, Not Patch

https://lukasniessen.medium.com/immutable-infrastructure-devops-why-you-should-replace-not-patch-e9a2cf71785e

https://redd.it/1nqg0i9
@r_devops

Why does every startup think they need to build their own incident management system?

Just joined a new company and they're super proud of their "custom incident response workflow" that's basically a Python noscript that creates Slack channels and a Notion page. Founder keeps talking about how "we're not like other companies, our incidents are different."

They're not different. Same dance every time service goes down, someone manually pages people, we all jump into a channel and start debugging while trying to remember if we updated the status page.

Previous engineer who built this thing left 6 months ago and nobody really understands how it works. Last week it created 15 incident channels for the same outage because of some edge case nobody thought of.

Every startup goes through this phase where they think incident management is their unique problem that needs a custom solution. Meanwhile we're burning engineering time maintaining this janky noscript instead of just buying something that works.

Anyone else dealt with this NIH syndrome around incident tooling? How do you convince leadership that some problems are worth paying someone else to solve?

https://redd.it/1nqigf2
@r_devops

The spam in this sub is unreal

Two posts today, sock puppet SEO accounts. Poster with a lame premise, commenter in to suggest a solution.

Cant remember what the first one was (they deleted their post), but the second was Atlassian - https://www.reddit.com/r/devops/s/M5DUQGRrtj

Mods, please take note and stop this nonsense.

https://redd.it/1nqm5v5
@r_devops

What certs/qualifications can I get as a Backend/DevOps to be more qualified and hirable?

hey, 23 year old male with a degree in CS I have a lot of experience that puts me in a really good place where I live I make 10 times more than what juniors make and I make 6-7 times what seniors make but I'm not good enough to get a sponsorship and go to a country that gives me decent livable money while I get more experiences so I can actually be something eventually

so the goal now is to get a job in North American, Australia, EU whatever just whatever country, I know if I go to the EU I will be making a lot less money that what I'm making now but it will be more than full time companies salary here and I will be finally able to advance my career and skills in an office job more than contracting

so what I need now it some advice, should I go into DevOps or focus on being a Backend dev? what certs or what should I do to make myself hirable? I need to leave here asap because its either slave salaries or no advancements in my career.

should I get a masters?

https://redd.it/1nqo56i
@r_devops

Need suggestions please

Hey everyone! I come from a non-IT background (5 years of experience at Amazon) and I've almost completed 90% of a DevOps course. My major concern now is resume creation. Also, once they see my relieving letter, my designation will be clearly visible. (I resigned 6 months ago due to personal reasons, and since then I've gained knowledge in DevOps. However, I did not work on any DevOps-related roles or services during my tenure.)

In addition, my CTC was comparatively lower and when they ask these questions, I'll be totally clueless. I'm no longer afraid of attending DevOpsinterviews since I feel confident, but these two points are worrying me. Any insights would be greatly helpful. Thank you.



https://redd.it/1nqpjgc
@r_devops

Exploring Terminals, TTYs, and PTYs

This post explores terminals, tty and pty.

https://cefboud.com/posts/terminals-pty-tty-pyte/

https://redd.it/1nqrujv
@r_devops

New Relic's CCU-based pricing is creating unpredictable costs, pushing teams to sample heavily

My teammate pointed out something about New Relic's pricing that I had to see for myself. They have this CCU (Compute Capacity Unit) pricing model that can lead to unpredictable costs.

When I went to their pricing page to check what he was talking about, I didn't even realize CCU-based and user-based are two separate pricing options. They present it in a way where it's easy to think CCU is just a component of their pricing, not a distinct model. Had to look twice to catch that.

I wrote about how their CCU pricing actually works based on our customer conversations. The model charges based on peak concurrent usage, so one traffic spike can blow up your monthly bill.

Has anyone here dealt with unexpected costs from CCU-based pricing? How do you handle capacity planning when your monitoring costs can spike unpredictably?

Look, as a competitor (I work at SigNoz), we're always analyzing what others are doing in the space. But this CCU pricing thing? I'm genuinely lost on how their customers budget for this.

https://redd.it/1nquexs
@r_devops

AI agent for internal documents

Hello there! As mentioned in the noscript, I want to create a chat that replies to people's questions using the internal documents. For the simplicity I've chosen open-webui, but the replies are quite slow. What have you used with good results? Thanks in advance!

https://redd.it/1nqvfj3
@r_devops

Do you ever just… vibe code the way I vibe write? 😅

So I’m not a developer, I’m in marketing. Which basically means half my life is staring at a blank doc at midnight, pouring coffee into my soul, and just writing whatever feels right in the moment. No plan. No brief. No strategy. Just vibes.

And then the next morning I’m like… “who the hell wrote this nonsense?” Oh right, me.

Lately I’ve been watching the devs I work with and holy sh*t, you guys do the exact same thing but with code. Someone gets a random 2am brainwave, spins up a repo, slaps code together like it’s jazz improv, and suddenly staging is broken and nobody wants to admit who touched it. That’s what I’ve now learned is called “vibe coding.”

As an outsider, it’s honestly hilarious and terrifying at the same time. Like bro, how is my chaotic draft doc somehow less dangerous than your chaotic repo? At least my typos don’t take prod down. I actually ended up writing a blog about this whole thing because I couldn’t stop laughing at the parallels… how marketers vibe write and devs vibe code in almost the same messy way. Dropped it here if anyone’s curious: https://www.codeant.ai/blogs/vibe-coding

But more importantly… I gotta know: do you all do this regularly? Also, let me know if you want me to cover more technical aspects in this blog for you next time.

https://redd.it/1nqxiz8
@r_devops

Is anyone else fighting the too many tools monster?

I swear half my job now is just… logging into things. We’ve got one tool for tickets, another for planning, another for infra as code changes, one more for approvals, then three different dashboards because nobody can agree which metrics actually matter.

At some point it stopped feeling like we were automating anything and started feeling like the tools were running us. Every new problem seems to spawn a new platform and before long we’re spending more time maintaining the toolchain than actually shipping.

Lately we’ve been questioning whether all this fragmentation is worth it. Would we actually move faster if we cut back and consolidated into fewer systems, even if they’re not best-in-class at every single thing? Or is that just wishful thinking and this kind of tool chaos is inevitable as you scale?

Did you double down on fewer tools and make them work harder? Or embrace the sprawl and just accept that integration glue is part of the job now?

https://redd.it/1nqz6hz
@r_devops

Docker container crashes on deploy-debug tips?

My Dockerized app keeps crashing right after deployment to a cloud server, but it runs fine locally-logs just show a vague "exit code 1" error. I’m using a Node.js image with a basic Dockerfile, and I’ve checked ports and env vars, but I’m stumped. Is it a memory issue, misconfigured cloud settings, or something in my container setup? Any go-to tools or tricks for pinpointing the cause?

https://redd.it/1nr1jjw
@r_devops