Armory CTO Isaac on G2K Best-Practices
6:36PM Jan 23, 2020
Hey it's radio care i'm here with Isaac our CTO and we're just going to talk a little bit about best practices around software delivery.
Yeah, so one of the things that happens as part of our installation into our customers environment is that we get a lot of metadata about how they're deploying pipelines, and so we can actually put our customers really into kind of a few groups so high performers medium performers low performers. And one of the things that we've been able to kind of pick up is, is that high performers can typically typically get 30 to 40 applications on to Spinnaker within nine months so this is a production ready rolling out into production, doing thousands of deployments a week, and those are those are typically our high performers and we can separate out those from the low performers that have nine months to a year only have one app in production. And because Spinnaker is typically tied with Kubernetes adoption. That's really kind of what we're driving to is Kubernetes adoption right or or cloud adoption. Right. And so, when, if you know what we've noticed is that when we look further into the data, the, the main discriminant here is actually culture. It's not the technology use it's not even the education of your own engineers in terms of like, what do they know or don't know, it's, it comes down to. Do you trust your engineers are your teams organized in small teams that they can make decisions on their own. Do you look at failure as a learning opportunity, or is it something that you absolutely are just trying to avoid at all costs so you essentially stifle innovation, because you want to avoid failure altogether. And when we, when we say failure, we don't mean like, like critical failure or even vocable harm failure, we are thinking of like just you know if there's a small outage, how is that considered, is that a learning opportunity for you to iterate and go back in retrospect, and make the pipelines or your code better or Is that considered terrible and we should never have that again and you don't have those retrospectives or learning moments. And so the more that these, these leaders become comfortable with the idea of autonomy and independence and trust of their teams and that they will learn and iterate better, the faster you will go. And, in fact, actually, you will result in having less errors and less failures in production because they're learning faster and your behaviors will change so the reason Netflix has chaos engineering is because their teams are independent from each other and what they say is like I cannot reliably depend on the services around me. What will I do in a failure scenario and let me make code to deal with that in a lot of these companies they don't have that because they make the, the opposite assumption is that all components should be working at all time, and their code never gets better and they never are resilient. And so, and they control everything so they move slower, they have less releases a lot less innovation and we can see that
in the data very very clearly there's an interesting getting Yang between the platform and the culture, where, when, failure means nuking the entire user base the organization doesn't have the psychological safety to treat failures learning and learning organization is going to treat failure as an opportunity to improve versus something to be avoided at all costs, and having a platform that allows for errors to be small blast radius to the limited problems to be fixed quickly with things like canaries in one click rollbacks and chaos engineering gives the company the space to have the psychological safety to then start to trust their developers more and give them more autonomy and freedom, so that they can learn faster. So it's a really interesting yin and yang between the people on the platform yeah
that's absolutely right. It's like where do you focus your energy, you focus in on Canarian and rollbacks, or do you focus it on kind of control from the top right, and we've seen that the more you kind of invest in the systems that have those guardrails and give you that psychological safety, the faster you move so absolutely right.
All right, well thank you Isaac for that quick that quick chat here and we'll also share some data about what Isaac's talking about and if you want to learn more, feel free to reach out to us. We typically work with global 2000 companies that are trying to digitally transform and usually move out of data centers and start to leverage the cloud, more more effectively. All right. Ciao.