 You want to hear 20 minutes on modern least privilege? Is that sound interesting? Okay. That's not too old school of a term to bring up. Awesome. I don't like spamming people. So I always do for a lot of thumbs up, thumbs down to make sure that people want to hear more words before I say them. So hopefully that'll go well. So I'm James and basically I've been rattling around the cloud native community or the modern application community for the better part of a decade and going on more now. And the experiences I want to share and where I'm helping drive the set of products at VMware called Tanzu, it really orients around the idea that we have a new paradigm around modern applications that introduces new ways of doing security, in addition to new ways of doing development operations. So I'm pretty excited about the overall term of DevSecOps. In some ways, it's not a perfect term, but it galvanizes this unification of these roles in a way that I think is important. And I've been with leaders of large organizations the days or weeks after major security breaches happen. And a lot of my thinking was heavily influenced. Sitting with the head of a bank with about an $8 billion technology budget, maybe a month after the Equifax breach happened. And he looked up at me in dinner, he's like, you know, if that happens to us, I have to sit in front of Congress. And he basically said, like, could we come up with a plan so that that would not be the default in my organization? Because it's not fun to run around tasking people constantly to turn the wheels. Like, how could we make modern security more of an affordance of an overall platformer system that I run as an organization and less of a potential? And a lot of what my passion in this space is around is the like, some of these modern security principles need not only be possible within the CNCF community, they need to become affordances. And I think that's sort of the distinction that I'm gonna push on and maybe some of my thoughts will be interesting. And so my ask and my call to action is like, if you agree with any of this, help make me smarter about it or help push it because I think it can just, you know, help us all in our day to day jobs. Does that sound interesting as a topic? Oh, I got more than one thumb, I'm gonna keep talking. Cool. If you use the term DevSecOps, I'm not really here to get in a big semantic debate if it's the cool thing. If we think of another way of making affordances, the security of first-class affordance of a platform, I'll use it, but that's what I got today. And I think part of the observation of modern applications was that they weren't quite as static. So I think if you think about the evolution of a platform like Instagram, it was sort of a failed startup that discovered this human interaction of applying a filter gave people the audacity to post their filters, their photos, maybe a hundred X more often. And that wasn't something you could sit down in a room with Accenture and write a spec out for an app. It was sort of an experimental emergent process. And I think that's my mental model when I think of modern applications is that they have emergent properties that require iteration. And so as all of these large organizations are going agile and discovering the business impact of emergent properties of applications, they're becoming more iterative. And that speed of iteration really to me is the big change from maybe the old model, old enough that I was alive in the 90s and some people today were not. And in the 90s, I started working at a big iron tech company called Sun. And every year we would deploy a server and you were in that very old school mode. In fact, my first day there was watching a customer do a six week consulting agreement to size the server. Six weeks to size the server. So I've really lived this evolution from large static process driven IT to more of this iterative emergent cloud pattern. But I do think that iterative emergent pattern is super interesting. And so I think just in the same way the applications changed, we can change security in this new world. And I'm gonna push on a couple dimensions of that. But I do like to study history a bit. And so as we're kind of suddenly there's this craze around, I was super excited to see all the secure supply chain work yesterday. It has its own conference now. DevSecOps is the hot term everywhere I go. And it's sort of like, wait a minute. If I read my computer history journals, it was 1974 that we were pretty clear that everything on a system should have as least privilege possible to execute its task. And in some ways, a lot of the systems we have today are much more wide open than that. And I think I traced that history back to the innovations of the web took a different turn than the early system designs which were largely backend systems for processes. They were much more collaboration systems. And so while the seventies, we were clear on least privilege. So anyway, I'm gonna go really fast. The founder of Ubiqui basically said, think about the web as a different paradigm on least privilege. It was a sharing first set of tech. Everything is about sharing fast. And I see DevSecOps as this really exciting new synthesis between least privilege system design of the early era and these collaboration technologies that we built like HTTP, which are originally just like, here's my document, everyone have it. Like how do we bring those two together as a modern ethos? That's really what drives me. An example of those two forces at play is the CNCF through some of the work Andreas has done has endorsed the cloud native build packs. And in the early days, it was sort of like, hey, everyone grab a Docker file and push it to prod, pull it from Docker hub and go. And I do think that things like build packs or a really nice engineering design of reducing the privilege of the system because by default now, everything is secured by default. So as opposed to saying to that developer, you need to always scan and do all of these things with your app by hand. Now you're saying there's a set of automations that introduce mechanistically privilege to the pipeline. And I do think paying attention to things like build packs is the first step. So here's some stuff I did in my previous generation of technology is more of a closed system called cloud foundry. But I worked with a lot of banks. One bank runs about 500,000 containers on this platform. And here's what they love about it is that there are fortances of the system which bake compliance into the developer experience, i.e. the developer has the surface area they need to get their job done and everything that they do by default is more secure. There's an ability to constantly repave the platform. So in the aftermath of Equifax, it was really a question of like, we can't wait 90 days to reintroduce, to rotate credentials, to repair things running in production and for frankly, for machines to expire. And so we introduced all of these things into this platform and banks by and large love it. And my passion right now is introducing that to the broader full Kubernetes community. We're baking into our developer experience because something called the Tanzu application platform which will do security scans, signing, et cetera by default. And come talk to me more about this, but we're baking some of these security processes right into the developer experience. And I think what's really interesting is that we're introducing, I think more of a decomposed set of templates that allow you to express the semantics that developers should see, what the app operations team should see and what core cluster admins should see in a way that allows you to decompose what I'll dares call a monolithic set of pipelines some people run into more like a microservice so that developer can reason about their surface area and a lot more things are secure signed by default. I think asking developers to reason about a 2000 function monolith is as unreasonable as asking them to reason about like a giant monolithic pipeline with no abstraction for them. That's my observation. And then we can bake security into these things. And the final note I'll leave you with is that an increasingly secure software supply chain doesn't eliminate runtime variance. And so you like, I know exactly when it went into prod and now it's running. So now what has it done since? And I think this is where the repaving modality is so important is that the ability now that you trust your software supply chain and you have GitOps functionality to constantly recreate it, why not use that? So why not make that a first class part of the system? And so what I want to do with Tanzu and we're increasingly doing with technologies like cluster API is introduce programmatic affordances to constantly repave the system to constantly repair the system as vulnerabilities come out and to constantly, and this will take some time rotate shared secrets, credentials, et cetera. And therefore I think we can, let me just close on this we can really bring together the two big trends that are happening now once you have a secure software supply chain and you have GitOps, why not reduce runtime variance and why not reduce bureaucratic overhead of getting patching done by constantly rebuilding that once you have systems like Knative which are excellent at doing first class routing between versions of applications and instances of applications, why not use that system? So that's my passion. That's what I'm working on. And my call to action is please help. Thank you. All right.