The following is from a Twitter thread by Gal Bashan of Epsagon and reprinted with permission. It has been edited for clarity when reading on a medium other than Twitter.
I chose this Twitter thread because it mirrors some things we have been talking about in this newsletter and does so more eloquently than I have been able to articulate before. Original thread is here.
Also, next week we'll be injecting some latency into this newsletter and arriving in your mailbox on Tuesday!
Serverless Advocates - We Are Doing Something Wrong
I have been putting this off for a long time but I really think it is important and has to be said. The trigger for this thread was a talk I gave at DevOpsDays Edinburgh. It wasn't the talk per-say, but the whole conference experience (which was awesome, but that's a topic for another thread).
Surrounded By Serverless
Here I am, with around 200 people who really care about the DevOps culture and practices - and so little of them has any experience with serverless tech. This blew my mind. I believe that people who embraced the DevOps culture can be the greatest advocates for serverless - there is a cultural fit with the shared responsibility model and you can’t get any leaner then serverless (CALMS anyone?).
What surprised me more was the lack of interest to learn about serverless - and that is not like the DevOps community, and I really think the fault is not with them but with us. So I tried to understand what are we doing wrong.
Kelsey Hightower already said it better than me, but I think we live in a serverless bubble. Jeremy Daly had an interesting tweet about this recently.
Serverless Bubble, On Both Sides
But the bubble is more problematic than it looks at first sight - because it is 2 sided.
On the one hand (the more obvious hand) when we hold serverless conferences most of the attendees are already sold on serverless. Now I am not saying we shouldn’t have them - community nurture is important - but we also need to focus on making the community more accessible.
And that’s where we get to the other hand (a.k.a my main point). On the other hand, because we are in the serverless bubble - we fail to see what’s going on outside of it. And that, I believe, is where we (or at least I) are truly wrong. We are not communicating our message right, because we don’t have a clear understanding of our customers. “No configuration needed” or “No servers to manage” is not a promising pitch to someone who adopted the DevOps culture, because DevOps is not about configuration nor servers.
DevOps is about making changes, measuring how they affected our work, repeat.
Now I really believe that serverless (not just Faas) IS what’s right for a lot of companies. But thinking is not enough. I have to show how going to serverless actually lowered my feature release time, reduced my mean time to bug resolution or improved any of my KPIs. Not just how I don’t have servers.
People (with me among them) say stuff like “I am tired of hearing people say serverless has servers. It doesn’t matter”. Guess what? It does matter. It matters because instead of trying to speak our customer's language - we ignore it when they try to speak ours. Since when ignoring a customers opinion openly became a valid advocating technique? If we really believe that serverless systems are the future - we have to do a better job at convincing our target audience. Just believing it is not enough.
I am not saying that the “serverless has server” thing is the biggest issue on our community's table. But it is just a small example of how we ignore our target demographic because we think we know better than them.
Convincing managers is not enough. It’s the dev and ops who have to deal with it eventually. And I think getting started is not accessible at all. Deploying a “Hello world!” Lambda != getting started. And currently getting started with a REAL serverless application (the right way) is somewhere between hard and undefined.
I sat with one of the guys at the conference. He said that for a manager serverless makes perfect sense, but for him, as a dev, it is just scary - and he is right. It has to be easier to get started (correctly) - much like on-boarding is important in any other service.
We need to talk about how existing companies can adopt serverless more easily. How they can integrate it with their existing application. We can’t build stuff that fits only applications which are completely serverless. I think that's not only naive but does not help adoption.
I know I am presenting this issue in a bit of an extreme manner, and Chris Munns has a point but that does not mean we don’t have to do our best to help adoption be as quick as possible.
My bottom line is that if we really believe in serverless, we have to start selling it better. We need to listen to our target demographic and adjust accordingly. I am not saying we are not already doing that, I just think we can do more.
That’s my 2 cents. I really think this is one of the biggest wrongs we have in the community. I am really interested in what the rest of you think about this.
I'm Tom from ServerlessOps and we provide a range of services around DevOps transformation, AWS migration, serverless training, and startup cloud operations. See our services to learn more.
Also let us know what you think of this week's Cold Start using the 👍/👎 links at the end of this email!
Tom at ServerlessOps