By R.D. Danes, a staff writer for SiliconANGLE
This article was written and produced as part of the Orate project, which helps organizations tell technical stories
Aren’t enterprises lucky the cloud-native ecosystem is growing so lavishly? A whole universe of well-honed tools is expanding before them. The means with which to modernize their infrastructure and applications are at their fingertips. The problem? They’re lost in a herd of tools that might work for some organizations, but not their own. At the Kubecon + CloudNativeCon North America 2018 conference in Seattle, Washington, last month, enterprise leaders gathered to share tips on scavenging the best tools from the bunch.
A panel titled, “Avoiding the Weeds in the Cloud Native Landscape,” brought together four professionals toiling in the enterprise-IT trenches. Moderator Priyanka Sharma, director of alliances for GitLab and contributor to the OpenTracing project, picked panelists’ brains for tricks to cloud-native tool shopping. They shared real-world successes and failures in procuring, trying, returning and swapping tools.
Why cloud native?
Two solid rules for IT teams wading into cloud native stood out in the discussion: 1) Do thorough research upfront, and 2) don’t hesitate to chuck a selection that doesn’t pan out.
Before embarking on a tool-hoarding spree to cure all your app-performance woes, make sure to have a reality check on what’s possible, advised Melissa Chapman, director of IT for PaaS and configuration management at CVS Health.
“If you have a bad piece of code or a bad application, and you think a tool is going to fix it, you are so wrong,” she said.
Enterprises need to keep their endgame top of mind. Why cloud native? Do they want to speed up application development and deployment? Give their developers more elbow room to innovate on impulse? Increase the agility of their DevOps teams? Do they know precisely how cloud-native technologies are going to help them achieve these goals?
“Tooling is a means to an end,” said Brendan Aye, director of platform architecture for T-Mobile.
“You don’t see Wall Street Journal articles about the great tool sets some companies put out there.”
With all the hype around fresh technology, it’s easy to forget that it’s useful only in as far as it improves business outcomes. What might work wonders for one business may flop fantastically in another. It might not be compatible older technologies, or staffers may lack the skills needed to work with it.
That said, picking useful tools isn’t entirely down to trial and error. Companies can certainly define goals and establish some standards that tools must meet before they give them a test run.
Introducing the unofficial AEIOU cloud-native standards body
Anyone remember learning the vowels?
Jasmine James, senior systems engineer for Delta Air Lines, and her colleagues repurposed the old “AEIOU” song. They turned it into an acronym and use it to size up cloud-native tools. It goes like this: Applicability, Enterprise-Ready, Integration, Overhead and Usefulness.
The most important element of all is usefulness, James said. Even if other businesses have had a ton of success with a tool, Delta won’t grab it until it’s sure it can get some juice out of it. “We want to make sure you can actually use them in the right way.”
Use-case cure for transformation anxiety attack
Usefulness is most easily determined on the basis of some well-defined use case. The business knows the problem, knows what hasn’t worked in the past and what might finally fix it.
An incremental approach to adopting cloud-native technologies and modernizing applications is much wiser than boiling the ocean, according to Matt Klein, software engineer for Lyft and creator of CNCF project Envoy.
“That’s really what I recommend to everyone on their journey is to really think about what problems are being faced and try to do them incrementally. Like, there’s no Big Bang solution,” he said. “Adopting a single technology is not going to transform a business overnight, but a single tool could vastly improve one use case or application. A series of simple steps like that could eventually result in a transformed business.”
Reel in, throw back, repeat for best catch
What if a tool doesn’t work? What if it’s a total disaster? Easy — throw it out and try something new.
“There’s dozens of tools for any single use case,” Aye said. “Perhaps CNCF’s project Prometheus 1.0 failed miserably in your company, but Prometheus 2.0 might deliver everything the first version lacked”.
Chapman agreed that rapid iteration is the royal road to finding the best fit in cloud-native tooling.
CVS wanted to digitize its receipts so that customers could view them on smartphones. It experimented with cloud-native technologies, failed, continued and finally succeeded.
“We started with some tools that we thought were going to help us and didn’t,” she said.
“And so we shifted very quickly — but it ended up being fantastic.”
To integrate or not to integrate?
When a company finally finds the tools that work for it, it may soon realize there are too many of them to manage. What then? Should it opt for tools that cluster features together?
“Perhaps not”, Aye said. “I don’t want to name vendors, but there are some logging tools now that started integrating metrics and some metrics tools that started integrating logging — and I feel like that’s a good way to be bad at both those things.”
Best-of-breed point solutions are usually preferable to mediocre medleys. The cloud-native ecosystem has work to do integrating tools for easier management without watering down their features.
We shall see how much progress it has made in May when the CNCF comes together again in Barcelona, Spain, for KubeCon + CloudNativeCon EU 2019 (call for proposals close January 19th).