Title: October 3, 2011 - A Day in the Life of DevOps and Chaos Engineering

---

Today was a day that felt like the tech world was on edge. Steve Jobs had just passed away, and while it’s clear his impact is huge, I couldn’t help but focus more on some of the other developments happening around us.

The Tools We Use: Chef vs Puppet

I’ve been wrestling with whether to stick with Chef or move over to Puppet for our configuration management needs. It’s like choosing between two favorite meals; both are good in their own ways, but I find myself leaning towards Chef because it’s more Ruby-based and easier for our team to grok.

One of the things that has really pushed me towards Chef is the ability to write custom resources and hooks. This allows us to handle complex tasks like setting up a database or managing multiple services without having to hack around in a language not designed for system administration tasks.

But let’s be real, Puppet fans are passionate, too. There’s a certain elegance to Puppet’s syntax and its declarative nature that can’t be denied. I think it’s the difference between driving an automatic versus a manual transmission - both get you where you need to go, but one feels more precise for some tasks.

Chaos Engineering at Netflix

Netflix has been doing chaos engineering for years now, and it makes me scratch my head wondering why everyone else isn’t doing it. We’ve had our fair share of outages over the past few months, and I keep thinking we need to start a little disruption party in our own environment. It’s like we’re all waiting for someone else to be the first to do it.

We have some rudimentary load testing and failover testing, but nothing that really simulates the kind of failure Netflix throws at itself. Maybe I’ll take the initiative and put together a small chaos engineering sandbox to test our systems in a real-world, noisy environment.

OpenStack: The Hype Machine

OpenStack is all the rage, and it’s hard not to get sucked into the hype cycle. But honestly, we’re still running on AWS for most of our infrastructure. I can see why companies are jumping on OpenStack; it offers great flexibility and a ton of open source features that can save us money in the long run. The problem is, we don’t have the resources to properly maintain and support an entire private cloud solution yet.

I’m hoping that as we grow, so will our budget for these kinds of initiatives. For now, AWS provides stability and ease-of-use that’s hard to match with OpenStack.

Heroku: A Part of Salesforce Now

Heroku was such a cool platform, but selling it to Salesforce is like someone buying your favorite band—you just can’t help feeling a bit weirded out about it. It was great knowing we could build complex web apps without worrying too much about infrastructure, and now that’s gone.

I’ve heard they’re moving towards more managed services, which might make sense given their acquisition. But part of me wonders if Heroku’s best days are behind us. I guess time will tell if Salesforce can make it work as a product.

NoSQL: The Next Big Thing?

NoSQL is everywhere you look, and everyone wants to dive in headfirst. We’ve been hesitant because we’re still trying to understand what problem it solves for us. There’s just so much complexity around data models that I’m not sure if the benefits outweigh the risks yet.

I might try out a couple of NoSQL databases on the side as proof-of-concepts, but until we see real use cases where they shine, I don’t think it’s worth disrupting our existing relational database setup.

Conclusion

Today felt like a mix of excitement and hesitation. Steve Jobs’ passing was a stark reminder of how quickly things can change in tech, and how much impact one person can have on the industry. As we move forward with tools like Chef and chaos engineering, I’m reminded that no matter what technologies come or go, it’s always about delivering value to our users.

Until next time,

Brandon

---