Cloud Native DevOps – Agile for Infrastructure

What best practices for agile infrastructure should you implement? Industry leaders gathered to discuss this topic at last Fall’s ONUG conference. Chris Swan of DXC Technology moderated a discussion with David McKay of InfluxData, Anne Currie of Container Solutions and Ben Hall of Katacoda. Below are their top five recommendations for enterprises today. Gather more insights by viewing their entire discussion here. 

Stay Up-to-Date

Technology is changing rapidly, especially DevOps. Swan asked each participant for tips on how they stay up-to-date with the latest trends. Here were their top six strategies.

  1. Newsletters: There are many tech newsletters that prove to be helpful. The participants specifically mentioned DevOps Weekly by Gareth Rushgrove of Snyk. 
  2. Conferences\Events: Both McKay and Hall emphasized the benefits of attending industry conferences, such as those hosted by ONUG. McKay speaks at a conference almost every week of the year. “It’s a great place to see what they are doing,” he said of the DevOps community. 
  3. Talk to Thought Leaders: That was Currie’s suggestion for getting your trends curated and neatly packaged for you. She prefers to simply invite thought leaders out for a coffee and ask them how their organization is doing. “Enterprises are good at telling you what they are doing,” Currie said. Take advantage of that. 
  4. Social Networking: McKay says he stays active on Twitter, following industry leaders to cut down on the unnecessary “noise.” 
  5. Read Trending Books: Cloud Native Attitude was one suggestion by Currie. 
  6. Podcasts: McKay also recommended subscribing to quality podcasts to gain insights in the industry. 

Sharpen Your Skills

The panel discussed the advantages of microlearning. Currie noted that the popularity of Microsoft certifications has fallen because people do not want to spend months of their lives learning a new skill. Instead, they want to quickly go in, test out a tool and move on. Microlearning platforms, such as Microsoft’s Learn and Google’s Qwiklabs, are providing that environment.

Swan referred to these platforms as “playpens,” allowing users to experiment and play with new tools. For example, Qwiklabs offers 90-minute challenges, somewhat like escape rooms, that give users a chance to solve a problem and earn badges while they do it. 

The group spent most of the time listening to Hall describe Katacoda, a highly successful microlearning platform. Hall says he founded Katacoda, realizing developers need a place to learn, experiment and figure out what tools will add value to their organization. Katacoda does that in a relatively short span of time. This process saves time and money because developers don’t get six months into a new process before realizing the tool is not a good fit. 

Hall calls his platform a “playground” for developers, a hands-on way of learning. Users can select topics on the homepage, such as “cloud native” or a specific programming language. Then, those topics are broken into problems. Developers can select the specific skill they need to develop, such as “how do I launch a container?” 

The “play” area launches a terminal where developers can see context, experiment with problem solving and learn fundamentals. If they break something, a simple “refresh” will reset the terminal. McKay has used Katacoda at InfluxData and describes it as “very interactive and gated at the same time. Fun. Definitely something I plan to do more often.” 

Make Going Green a Priority

Currie mostly led this discussion, emphasizing that companies must make hosting in a green location part of their procurement process. She predicts that in five years all tech centers will be required to be carbon neutral. Companies need to make it a priority now because it will get increasingly hard to make the shift down the road.

Google Cloud is really leading the way. They are already 100 percent renewable. AWS only has a few green regions. Their goal is to be completely green by 2030. However, they could go faster with more customer demand. Currie encouraged IT professionals to not underestimate the impact of their voice. Tech has just as large of a carbon footprint as the aviation industry, and we need to call for change. 

Going forward, “we’re going to need to know the carbon footprint of our products and have a story on what we’ve done to make them more efficient,” Currie predicted. If AWS moves up their date, it would be a huge win for so many. That will only happen if customers demand it. Start making “going green” part of your criteria when choosing cloud hosting services.

“Talk to your Amazon Rep,” Curried suggested. Let them know going green is important to you. Right now, Amazon only has four carbon neutral regions, Dublin, Frankfort, Canada and Oregon. If you are not in one of those four regions, about half of your power for hosting is from fossil fuels. Let them know you care!

McKay affirmed that he doesn’t see “going green” as part of the procurement conversation, but it needs to be. Hall added that Katacoda recently moved its data centers to Finland because of the lower carbon footprint, as well as the cost savings. He noted that hybrid cloud is enabling companies to be flexible in choosing their drivers. Bump “going green” to the top of your priority list.

Shift from Monitoring to Observability 

McKay focused on the trend from standard metric-based monitoring to observability. Cloud has made it necessary to change the way we monitor systems. A decade ago, we simply wrote monolithic applications that ran on a host. We would check to see if that machine was online and if the port was open. This was the start of metric-based monitoring. A metric or check would indicate a problem, and we would treat the symptom. 

Our shift to cloud requires us to look at the system as a whole. With metric-based monitoring, we never really addressed the cause of a problem. This is where observability has changed the game. Today we must observe a complex structure that includes applications, networks, infrastructures, traffic and software. “Observability helps us understand these systems,” says McKay. “It increases visibility.” Observability involves taking metrics, logs and traces to understand the system as a whole. This approach creates as much context as possible. 

Currie agreed with McKay, saying that observability is “utterly necessary.” She emphasized shifting our thinking to that of figuring out what kinds of diagnostics can be put into products to make them more observable. She asked McKay what kinds of factors he would put into making these decisions. McKay recommended dynamic sampling, the process of collecting samples at “the edge,” and keeping a memory representation of the data you have. “There’s no centralized sampling I’ve seen work,” says McKay. “The only way to do it is to push it to the edge.” 

Automate Security

Panel participants agreed that automating security is necessary. “Attackers are automated, and we must be too,” warned Currie. She emphasized that we must patch constantly. It’s the only strategy that works because we are all working with open source modules that come from many different places. Unfortunately, companies can not keep up with manual patches. Automation is a must. 

Ben also hit on the importance of automation. He encouraged listeners to use services available within their applications that will proactively scan the application against the latest disclosure. 

Access more valuable resources at or sign up for our Fall 2020 conference. 

Author's Bio

Guest Author

Guest Author