(upbeat music) >> Hello everyone, my name is Dave Vellante, and we're digging into the many facets of the software supply chain and how to better manage digital risk. I'd like to introduce Kirsten Newcomer, who is the Director of Cloud and DevSecOps Strategy at Red Hat. Hello Kirsten, welcome. >> Hello Dave, great to be here with you today. >> Let's dive right in. What technologies and practices should we be thinking about that can help improve the security posture within the software supply chain? >> So I think the most important thing for folks to think about really is adopting DevSecOps. And while organizations talk about DevSecOps, and many folks have adopted DevOps, they tend to forget the security part of DevSecOps. And so for me, DevSecOps is both DevSec, how do I shift security left into my supply chain, and SecOps which is a better understood and more common piece of the puzzle, but then closing that loop between what issues are discovered in production and feeding that back to the development team to ensure that we're really addressing that supply chain. >> Yeah I heard a stat. I don't know what the source is, I don't know if it's true, but it probably is that around 50% of the organizations in North America, don't even have a SecOps team. Now of course that probably includes a lot of smaller organizations, but the SecOps team, they're not doing DevSecOps, but so what are organizations doing for supply chain security today? >> Yeah, I think the most common practice, that people have adopted is vulnerability scanning. And so they will do that as part of their development process. They might do it at one particular point, they might do it at more than one point. But one of the challenges that, we see first of all, is that, that's the only security gate that they've integrated into their supply chain, into their pipeline. So they may be scanning code that they get externally, they may be scanning their own code. But the second challenge is that the results take so much work to triage. This is static vulnerability scanning. You get information that is not in full context, because you don't know whether a vulnerability is truly exploitable, unless you know how exposed that particular part of the code is to the internet, for example, or to other aspects. And so it's just a real challenge for organizations, who are only looking at static vulnerability data, to figure out what the right steps to take are to manage those. And there's no way we're going to wind up with zero vulnerabilities, in the code that we're all working with today. Things just move too quickly. >> Is that idea of vulnerability scanning, is it almost like sampling where you may or may not find the weakest link? >> I would say that it's more comprehensive than that. The vulnerability scanners that are available, are generally pretty strong, but they are, again, if it's a static environment, a lot of them rely on NVD database, which typically it's going to give you the worst case scenario, and by nature can't account for things like, was the software that you're scanning built with controls, mitigations built in. It's just going to tell you, this is the package, and this is the known vulnerabilities associated with that package. It's not going to tell you whether there were compiler time flags, that may be mitigated that vulnerability. And so it's almost overwhelming for organizations, to prioritize that information, and really understand it in context. And so when I think about the closed loop feedback, you really want not just that static scan, but also analysis that takes into account, the configuration of the application, and the runtime environment and any mitigations that might be present there. >> I see, thank you for that. So, given that this digital risk and software supply chains are now front and center, we read about them all the time now, how do you think organizations are responding? What's the future of software supply chain going to look like? >> That's a great one. So I think organizations are scrambling. We've certainly at Red Hat, We've seen an increase in questions, about Red Hat's own supply chain security, and we've got lots of information that we can share and make available. But I think also we're starting to see, this strong increased interest, in security bill of materials. So I actually started working with, automation and standards around security bill of materials, a number of years ago. I participated in The Linux Foundation, SPDX project. There are other projects like CycloneDX. But I think all organizations are going to need to, those of us who deliver software, we're going to need to provide S-bombs and consumers of our software should be looking for S-bombs, to help them understand, to build transparency across the projects. And to facilitate that automation, you can leverage the data, in a software package list, to get a quick view of vulnerabilities. Again, you don't have that runtime context yet, but it saves you that step, perhaps of having to do the initial scanning. And then there are additional things that folks are looking at. Attested pipelines is going to be key, for building your custom software. As you pull the code in and your developers build their solutions, their applications, being able to vet the steps in your pipeline, and attest that nothing has happened in that pipeline, is really going to be key. >> So the software bill of materials is going to give you, a granular picture of your software, and then what the chain of, providence if you will or? >> Well, an S-bomb depending on the format, an S-bomb absolutely can provide a chain of providence. But another thing when we think about it, from the security angles, so there's the providence, where did this come from? Who provided it to me? But also with that bill of materials, that list of packages, you can leverage tooling, that will give you information about vulnerability information about those packages. At Red Hat we don't think that vulnerability info should be included in the S-bomb, because vulnerability data changes everyday. But, it saves you a step potentially. Then you don't necessarily have to be so concerned about doing the scan, you can pull data about known vulnerabilities for those packages without a scan. Similarly the attestation in the pipeline, that's about things like ensuring that, the code that you pull into your pipeline is signed. Signatures are in many ways of more important piece for defining providence and getting trust. >> Got it. So I was talking to Asiso the other day, and was asking her okay, what are your main challenges, kind of the standard analyst questions, if you will. She said look, I got great people, but I just don't have enough depth of talent, to handle, the challenges I'm always sort of playing catch up. That leads one to the conclusion, okay, automation is potentially an answer to address that problem, but the same time, people have said to me, sometimes we put too much faith in automation. some say okay, hey Kirsten help me square the circle. I want to automate because I lack the talent, but it's not, it's not sufficient. What are your thoughts on automation? >> So I think in the world we're in today, especially with cloud native applications, you can't manage without automation, because things are moving too quickly. So I think the way that you assess whether automation is meeting your goals becomes critical. And so looking for external guidance, such as the NIST's Secure Software Development Framework, that can help. But again, when we come back, I think, look for an opinionated position from the vendors, from the folks you're working with, from your advisors, on what are the appropriate set of gates. And we've talked about vulnerability scanning, but analyzing the configed data for your apps it's just as important. And so I think we have to work together as an industry, to figure out what are the key security gates, how do we audit the automation, so that I can validate that automation and be comfortable, that it is actually meeting the needs. But I don't see how we move forward without automation. >> Excellent. Thank you. We were forced into digital, without a lot of thought. Some folks, it's a spectrum, some organizations are better shape than others, but many had to just dive right in without a lot of strategy. And now people have sat back and said, okay, let's be more planful, more thoughtful. So as you, and then of course, you've got, the supply chain hacks, et cetera. How do you think the whole narrative and the strategy is going to change? How should it change the way in which we create, maintain, consume softwares as both organizations and individuals? >> Yeah. So again, I think there's going to be, and there's already, need request for more transparency, from software vendors. This is a place where S-bombs play a role, but there's also a lot of conversation out there about zero trust. So what does that mean in, you have to have a relationship with your vendor, that provides transparency, so that you can assess the level of trust. You also have to, in your organization, determine to your point earlier about people with skills and automation. How do you trust, but verify? This is not just with your vendor, but also with your internal supply chain. So trust and verify remains key. That's been a concept that's been around for a while. Cloud native doesn't change that, but it may change the tools that we use. And we may also decide what are our trust boundaries. Are they where are we comfortable trusting? Where do we think that zero trust is more applicable place, a more applicable frame to apply? But I do think back to the automation piece, and again, it is hard for everybody to keep up. I think we have to break down silos, we have to ensure that teams are talking across those silos, so that we can leverage each other's skills. And we need to think about managing everything as code. What I like about the everything is code including security, is it does create auditability in new ways. If you're managing your infrastructure, and get Ops like approach your security policies, with a get Ops like approach, it provides visibility and auditability, and it enables your dev team to participate in new ways. >> So when you're talking about zero trust I think, okay, I can't trust users, I got to trust the verified users, machines, employees, my software, my partners. >> Yap >> Every possible connection point. >> Absolutely. And this is where both attestation and identity become key. So being able to, I mean, the SolarWinds team has done a really interesting set of things with their supply chain, after they were, in response to the hack they were dealing with. They're now using Tekton CD chains, to ensure that they have, attested every step in their supply chain process, and that they can replicate that with automation. So they're doing a combination of, yep. We've got humans who need to interact with the chain, and then we can validate every step in that chain. And then workload identity, is a key thing for us to think about too. So how do we assert identity for the workloads that are being deployed to the cloud and verify whether that's with SPIFFE SPIRE, or related projects verify, that the workload is the one that we meant to deploy and also runtime behavioral analysis. I know we've been talking about supply chain, but again, I think we have to do this closed loop. You can't just think about shifting security left. And I know you mentioned earlier, a lot of teams don't have SecOps, but there are solutions available, that help assess the behavior and runtime, and that information can be fed back to the app dev team, to help them adjust and verify and validate. Where do I need to tighten my security? >> Am glad you brought up the SolarWinds to Kirsten what they're doing. And as I remember after 911, everyone was afraid to fly, but it was probably the safest time in history to fly. And so same analogy here. SolarWinds probably has learned more about this and its reputation took a huge hit. But if you had to compare, what SolarWinds has learned and applied, at the speed at which they've done it with maybe, some other software suppliers, you might find that they've actually done a better job. It's just, unfortunately, that something hit that we never saw before. To me it was Stuxnet, like we'd never seen anything like this before, and then boom, we've entered a whole new era. I'll give you the last word Kirsten. >> No just to agree with you. And I think, again, as an industry, it's pushed us all to think harder and more carefully about where do we need to improve? What tools do we need to build to help ourselves? Again, S-bombs have been around, for a good 10 years or so, but they are enjoying a resurgence of importance signing, image signing, manifest signing. That's been around for ages, but we haven't made it easy to integrate that into the supply chain, and that's work that's happening today. Similarly that attestation of a supply chain, of a pipeline that's happening. So I think as a industry, we've all recognized, that we need to step up, and there's a lot of creative energy going into improving in this space. >> Excellent Kirsten Newcomer, thanks so much for your perspectives. Excellent conversation. >> My pleasure, thanks so much. >> You're welcome. And you're watching theCUBE, the leader in tech coverage. (soft music)

(upbeat music) >> Hello, and welcome back to AWS Startup Showcase, I'm John Furrier, your host. This is the Hero panel, the AWS Heroes. These are folks that have a lot of experience in Open Source, having fun building great projects and commercializing the value and best practices of Open Source innovation. We've got some great guests here. Liz Rice, Chief Open Source Officer, Isovalent. CUBE alumni, great to see you. Brian LeRoux, who is the Co-founder and CTO of begin.com. Erica Windisch who's an Architect for Developer Experience. AWS Hero, also CUBE alumni. Casey Lee, CTO Gaggle. Doing some great stuff in ed tech. Great collection of experts and experienced folks doing some fun stuff, welcome to this conversation this CUBE panel. >> Hi. >> Thanks for having us. >> Hello. >> Let's go down the line. >> I don't normally do this, but since we're remote and we have such great guests, go down the line and talk about why Open Source is important to you guys. What projects are you currently working on? And what's the coolest thing going on there? Liz we'll start with you. >> Okay, so I am very involved in the world of Cloud Native. I'm the chair of the technical oversight committee for the Cloud Native Computing Foundation. So that means I get to see a lot of what's going on across a very broad range of Cloud Native projects. More specifically, Isovalent. I focus on Cilium, which is it's based on a technology called EBPF. That is to me, probably the most exciting technology right now. And then finally, I'm also involved in an organization called OpenUK, which is really pushing for more use of open technologies here in the United Kingdom. So spread around lots of different projects. And I'm in a really fortunate position, I think, to see what's happening with lots of projects and also the commercialization of lots of projects. >> Awesome, Brian what project are you working on? >> Working project these days called Architect. It's a Open Source project built on top of AWSM. It adds a lot of sugar and terseness to the SM experience and just makes it a lot easier to work with and get started. AWS can be a little bit intimidating to people at times. And the Open Source community is stepping up to make some of that bond ramp a little bit easier. And I'm also an Apache member. And so I keep a hairy eyeball on what's going on in that reality all the time. And I've been doing this open-source thing for quite a while, and yeah, I love it. It's a great thing. It's real science. We get to verify each other's work and we get to expand and build on human knowledge. So that's a huge honor to just even be able to do that and I feel stoked to be here so thanks for having me. >> Awesome, yeah, and totally great. Erica, what's your current situation going on here? What's happening? >> Sure, so I am currently working on developer experience of a number of Open Source STKS and CLI components from my current employer. And previously, recently I left New Relic where I was working on integrating with OpenTelemetry, as well as a number of other things. Before that I was a maintainer of Docker and of OpenStack. So I've been in this game for a while as well. And I tend to just put my fingers in a lot of little pies anywhere from DVD players 20 years ago to a lot of this open telemetry and monitoring and various STKs and developer tools is where like Docker and OpenStack and the STKs that I work on now, all very much focusing on developer as the user. >> Yeah, you're always on the wave, Erica great stuff. Casey, what's going on? Do you got some great ed techs happening? What's happening with you? >> Yeah, sure. The primary Open Source project that I'm contributing to right now is ACT. This is a tool I created a couple of years back when GitHub Actions first came out, and my motivation there was I'm just impatient. And that whole commit, push, wait time where you're testing out your pipelines is painful. And so I wanted to build a tool that allowed developers to test out their GitHub Actions workflows locally. And so this tool uses Docker containers to emulate, to get up action environment and gives you fast feedback on those workflows that you're building. Lot of innovation happening at GitHub. And so we're just trying to keep up and continue to replicate those new features functionalities in the local runner. And the biggest challenge I've had with this project is just keeping up with the community. We just passed 20,000 stars, and it'd be it's a normal week to get like 10 PRs. So super excited to announce just yesterday, actually I invited four of the most active contributors to help me with maintaining the project. And so this is like a big deal for me, letting the project go and bringing other people in to help lead it. So, yeah, huge shout out to those folks that have been helping with driving that project. So looking forward to what's next for it. >> Great, we'll make sure the SiliconANGLE riders catch that quote there. Great call out. Let's start, Brian, you made me realize when you mentioned Apache and then you've been watching all the stuff going on, it brings up the question of the evolution of Open Source, and the commercialization trends have been very interesting these days. You're seeing CloudScale really impact also with the growth of code. And Liz, if you remember, the Linux Foundation keeps making projections and they keep blowing past them every year on more and more code and more and more entrance coming in, not just individuals, corporations. So you starting to see Netflix donates something, you got Lyft donate some stuff, becomes a project company forms around it. There's a lot of entrepreneurial activity that's creating this new abstraction layers, new platforms, not just tools. So you start to see a new kickup trajectory with Open Source. You guys want to comment on this because this is going to impact how fast the enterprise will see value here. >> I think a really great example of that is a project called Backstage that's just come out of Spotify. And it's going through the incubation process at the CNCF. And that's why it's front of mind for me right now, 'cause I've been working on the due diligence for that. And the reason why I thought it was interesting in relation to your question is it's spun out of Spotify. It's fully Open Source. They have a ton of different enterprises using it as this developer portal, but they're starting to see some startups emerging offering like a hosted managed version of Backstage or offering services around Backstage or offering commercial plugins into Backstage. And I think it's really fascinating to see those ecosystems building up around a project and different ways that people can. I'm a big believer. You cannot sell the Open Source code, but you can sell other things that create value around Open Source projects. So that's really exciting to see. >> Great point. Anyone else want to weigh in and react to that? Because it's the new model. It's not the old way. I mean, I remember when I was in college, we had the Pirate software. Open Source wasn't around. So you had to deal under the table. Now it's free. But I mean the old way was you had to convince the enterprise, like you've got a hard knit, it builds the community and the community manage the quality of the code. And then you had to build the company to make sure they could support it. Now the companies are actually involved in it, right? And then new startups are forming faster. And the proof points are shorter and highly accelerated for that. I mean, it's a whole new- >> It's a Cambrian explosion, and it's great. It's one of those things that it's challenging for the new developers because they come in and they're like, "Whoa, what is all this stuff that I'm supposed to figure out?" And there's no right answer and there's no wrong answer. There's just tons of it. And I think that there's a desire for us to have one sort of well-known trot and happy path, that audience we're a lot better with a more diverse community, with lots of options, with lots of ways to approach these problems. And I think it's just great. A challenge that we have with all these options and all these Cambrian explosion of projects and all these competing ideas, right now, the sustainability, it's a bit of a tricky question to answer. We know that there's a commercialization aspect that helps us fund these projects, but how we compose the open versus the commercial source is still a bit of a tricky question and a tough one for a lot of folks. >> Erica, would you chime in on that for a second. I want to get your angle on that, this experience and all this code, and I'm a new person, I'm an existing person. Do I get like a blue check mark and verify? I mean, these are questions like, well, how do you navigate? >> Yeah, I think this has been something happening for a while. I mean, back in the early OpenStack days, 2010, for instance, Rackspace Open Sourcing, OpenStack and ANSU Labs and so forth, and then trying, having all these companies forming in creating startups around this. I started at a company called Cloudccaling back in late 2010, and we had some competitors such as Piston and so forth where a lot of the ANSUL Labs people went. But then, the real winners, I think from OpenStack ended up being the enterprises that jumped in. We had Red Hat in particular, as well as HP and IBM jumping in and investing in OpenStack, and really proving out a lot of... not that it was the first time, but this is when we started seeing billions of dollars pouring into Open Source projects and Open Source Foundations, such as the OpenStack Foundation, which proceeded a lot of the things that we now see with the Linux Foundation, which was then created a little bit later. And at the same time, I'm also reflecting a little bit what Brian said because there are projects that don't get funded, that don't get the same attention, but they're also getting used quite significantly. Things like Log4j really bringing this to the spotlight in terms of projects that are used everywhere by everything with significant outsized impacts on the industry that are not getting funded, that aren't flashy enough, that aren't exciting enough because it's just logging, but a vulnerability in it brings every everything and everybody down and has possibly billions of dollars of impact to our industry because nobody wanted to fund this project. >> I think that brings up the commercialization point about maybe bringing a venture capital model in saying, "Hey, that boring little logging thing could be a key ingredient for say solving some observability problems so I think let's put some cash." Again then we'd never seen that before. Now you're starting to see that kind of a real smart investment thesis going into Open Source projects. I mean, Promethease, Crafter, these are projects that turned off companies. This is turning up companies. >> A decade ago, there was no money in Dev tools that I think that's been fully debunked now. They used to be a concept that the venture community believed, but there's just too much evidence to the contrary, the companies like Cash Court, Datadog, the list goes on and on. I think the challenge for the Open Source (indistinct) comes back to foundations and working (indistinct) these developers make this code safe and secure. >> Casey, what's your reaction to all of this? You've got, so a project has gained some traction, got some momentum. There's a lot of mission critical. I won't say white spaces, but the opportunities in the big cloud game happening. And there's a lot of, I won't say too many entrepreneurial, but there's a lot of community action happening that's precommercialization that's getting traction. How does this all develop naturally and then vector in quickly when it hits? >> Yeah, I want to go back to the Log4j topic real quick. I think that it's a great example of an area that we need to do better at. And there was a cool article that Rob Pike wrote describing how to quantify the criticality. I think that's sort of quantifying criticality was the article he wrote on how to use metrics, to determine how valuable, how important a piece of Open Source is to the community. And we really need to highlight that more. We need a way to make it more clear how important this software is, how many people depend on it and how many people are contributing to it. And because right now we all do that. Like if I'm going to evaluate an Open Source software, sure, I'll look at how many stars it has and how many contributors it has. But I got to go through and do all that work myself and come up with. It would be really great if we had an agreed upon method for ranking the criticality of software, but then also the risk, hey, that this is used by a ton of people, but nobody's contributing to it anymore. That's a concern. And that would be great to potential users of that to signal whether or not it makes sense. The Open Source Security Foundation, just getting off the ground, they're doing some work in this space, and I'm really excited to see where they go with that looking at ways to stop score critically. >> Well, this brings up a good point while we've got everyone here, let's take a plug and plug a project you think that's not getting the visibility it needs. Let's go through each of you, point out a project that you think people should be looking at and talking about that might get some free visibility here. Anyone want to highlight projects they think should be focused more on, or that needs a little bit of love? >> I think, I mean, particularly if we're talking about these sort of vulnerability issues, there's a ton of work going on, like in the Secure Software Foundation, other foundations, I think there's work going on in Apache somewhere as well around the bill of material, the software bill of materials, the Secure Software supply chain security, even enumerating your dependencies is not trivial today. So I think there's going to be a ton of people doing really good work on that, as well as the criticality aspect. It's all like that. There's a really great xkcd cartoon with your software project and some really big monolithic lumps. And then, this tiny little piece in a very important point that's maintained by somebody in his bedroom in Montana or something and if you called it out. >> Yeah, you just opened where the next lightening and a bottle comes from. And this is I think the beauty of Open Source is that you get a little collaboration, you get three feet in a cloud of dust going and you get some momentum, and if it's relevant, it rises to the top. I think that's the collective intelligence of Open Source. The question I want to ask that the panel here is when you go into an enterprise, and now that the game is changing with a much more collaborative and involved, what's the story if they say, hey, what's in it for me, how do I manage the Open Source? What's the current best practice? Because there's no doubt I can't ignore it. It's in everything we do. How do I organize around it? How do I build around it to be more efficient and more productive and reduce the risk on vulnerabilities to managing staff, making sure the right teams in place, the right agility and all those things? >> You called it, they got to get skin in the game. They need to be active and involved and donating to a sustainable Open Source project is a great way to start. But if you really want to be active, then you should be committing. You should have a goal for your organization to be contributing back to that project. Maybe not committing code, it could be committing resources into the darks or in the tests, or even tweeting about an Open Source project is contributing to it. And I think a lot of these enterprises could benefit a lot from getting more active with the Open Source Foundations that are out there. >> Liz, you've been actively involved. I know we've talked personally when the CNCF started, which had a great commercial uptake from companies. What do you think the current state-of-the-art kind of equation is has it changed a little bit? Or is it the game still the same? >> Yeah, and in the early days of the CNCF, it was very much dominated by vendors behind the project. And now we're seeing more and more membership from end-user companies, the kind of enterprises that are building their businesses on Cloud Native, but their business is not in itself. That's not there. The infrastructure is not their business. And I think seeing those companies, putting money in, putting time in, as Brian says contributing resources quite often, there's enough money, but finding the talent to do the work and finding people who are prepared to actually chop the wood and carry the water, >> Exactly. >> that it's hard. >> And if enterprises can find peoples to spend time on Open Source projects, help with those chores, it's hugely valuable. And it's one of those the rising tide floats all the boats. We can raise security, we can reduce the amount of dependency on maintain projects collectively. >> I think the business models there, I think one of the things I'll react to and then get your guys' comments is remember which CubeCon it was, it was one of the early ones. And I remember seeing Apple having a booth, but nobody was manning. It was just an Apple booth. They weren't doing anything, but they were recruiting. And I think you saw the transition of a business model where the worry about a big vendor taking over a project and having undue influence over it goes away because I think this idea of participation is also talent, but also committing that talent back into the communities as a model, as a business model, like, okay, hire some great people, but listen, don't screw up the Open Source piece of it 'cause that's a critical. >> Also hire a channel, right? They can use those contributions to source that talent and build the reputation in the communities that they depend on. And so there's really a lot of benefit to the larger organizations that can do this. They'll have a huge pipeline of really qualified engineers right out the gate without having to resort to cheesy whiteboard interviews, which is pretty great. >> Yeah, I agree with a lot of this. One of my concerns is that a lot of these corporations tend to focus very narrowly on certain projects, which they feel that they depend greatly, they'll invest in OpenStack, they'll invest in Docker, they'll invest in some of the CNCF projects. And then these other projects get ignored. Something that I've been a proponent of for a little bit for a while is observability of your dependencies. And I don't think there's quite enough projects and solutions to this. And it sounds maybe from lists, there are some projects that I don't know about, but I also know that there's some startups like Snyk and so forth that help with a little bit of this problem, but I think we need more focus on some of these edges. And I think companies need to do better, both in providing, having some sort of solution for observability of the dependencies, as well as understanding those dependencies and managing them. I've seen companies for instance, depending on software that they actively don't want to use based on a certain criteria that they already set projects, like they'll set a requirement that any project that they use has a code of conduct, but they'll then use projects that don't have codes of conduct. And if they don't have a code of conduct, then employees are prohibited from working on those projects. So you've locked yourself into a place where you're depending on software that you have instructed, your employees are not allowed to contribute to, for certain legal and other reasons. So you need to draw a line in the sand and then recognize that those projects are ones that you don't want to consume, and then not use them, and have observability around these things. >> That's a great point. I think we have 10 minutes left. I want to just shift to a topic that I think is relevant. And that is as Open Source software, software, people develop software, you see under the hood kind of software, SREs developing very quickly in the CloudScale, but also you've got your classic software developers who were writing code. So you have supply chain, software supply chain challenges. You mentioned developer experience around how to code. You have now automation in place. So you've got the development of all these things that are happening. Like I just want to write software. Some people want to get and do infrastructure as code so DevSecOps is here. So how does that look like going forward? How has the future of Open Source going to make the developers just want to code quickly? And the folks who want to tweak the infrastructure a bit more efficient, any views on that? >> At Gaggle, we're using AWS' CDK, exclusively for our infrastructure as code. And it's a great transition for developers instead of writing Yammel or Jason, or even HCL for their infrastructure code, now they're writing code in the language that they're used to Python or JavaScript, and what that's providing is an easier transition for developers into that Infrastructure as code at Gaggle here, but it's also providing an opportunity to provide reusable constructs that some Devs can build on. So if we've got a very opinionated way to deploy a serverless app in a database and do auto-scaling behind and all stuff, we can present that to a developer as a library, and they can just consume it as it is. Maybe that's as deep as they want to go and they're happy with that. But then they want to go deeper into it, they can either use some of the lower level constructs or create PRs to the platform team to have those constructs changed to fit their needs. So it provides a nice on-ramp developers to use the tools and languages they're used to, and then also go deeper as they need. >> That's awesome. Does that mean they're not full stack developers anymore that they're half stack developers they're taking care of for them? >> I don't know either. >> We'll in. >> No, only kidding. Anyway, any other reactions to this whole? I just want to code, make it easy for me, and some people want to get down and dirty under the hood. >> So I think that for me, Docker was always a key part of this. I don't know when DevSecOps was coined exactly, but I was talking with people about it back in 2012. And when I joined Docker, it was a part of that vision for me, was that Docker was applying these security principles by default for your application. It wasn't, I mean, yes, everybody adopted because of the portability and the acceleration of development, but it was for me, the fact that it was limiting what you could do from a security angle by default, and then giving you these tuna balls that you can control it further. You asked about a project that may not get enough recognition is something called DockerSlim, which is designed to optimize your containers and will make them smaller, but it also constraints the security footprint, and we'll remove capabilities from the container. It will help you build security profiles for app armor and the Red Hat one. SELinux. >> SELinux. >> Yeah, and this is something that I think a lot of developers, it's kind of outside of the realm of things that they're really thinking about. So the more that we can automate those processes and make it easier out of the box for users or for... when I say users, I mean, developers, so that it's straightforward and automatic and also giving them the capability of refining it and tuning it as needed, or simply choosing platforms like serverless offerings, which have these security constraints built in out of the box and sometimes maybe less tuneable, but very strong by default. And I think that's a good place for us to be is where we just enforced these things and make you do things in a secure way. >> Yeah, I'm a huge fan of Kubernetes, but it's not the right hammer for every nail. And there are absolutely tons of applications that are better served by something like Lambda where a lot more of that security surface is taken care of for the developer. And I think we will see better tooling around security profiling and making it easier to shrink wrap your applications that there are plenty of products out there that can help you with this in a cloud native environment. But I think for the smaller developer let's say, or an earlier stage company, yeah, it needs to be so much more straightforward. Really does. >> Really an interesting time, 10 years ago, when I was working at Adobe, we used to requisition all these analysts to tell us how many developers there were for the market. And we thought there was about 20 million developers. If GitHub's to be believed, we think there is now around 80 million developers. So both these groups are probably wrong in their numbers, but the takeaway here for me is that we've got a lot of new developers and a lot of these new developers are really struck by a paradox of choice. And they're typically starting on the front end. And so there's a lot of movement in the stack moved towards the front end. We saw that at re:Invent when Amazon was really pushing Amplify 'cause they're seeing this too. It's interesting because this is where folks start. And so a lot of the obstructions are moving in that direction, but maybe not always necessarily totally appropriate. And so finding the right balance for folks is still a work in progress. Like Lambda is a great example. It lets me focus totally on just business logic. I don't have to think about infrastructure pretty much at all. And if I'm newer to the industry, that makes a lot of sense to me. As use cases expand, all of a sudden, reality intervenes, and it might not be appropriate for everything. And so figuring out what those edges are, is still the challenge, I think. >> All right, thank you very much for coming on the CUBE here panel. AWS Heroes, thanks everyone for coming. I really appreciate it, thank you. >> Thank you. >> Thank you. >> Okay. >> Thanks for having me. >> Okay, that's a wrap here back to the program and the awesome startups. Thanks for watching. (upbeat music)

(upbeat music) >> Hello everyone, my name is Dave Vellante, and we're digging into the many facets of the software supply chain and how to better manage digital risk. I'd like to introduce Kirsten Newcomer, who is the Director of Cloud and DevSecOps Strategy at Red Hat. Hello Kirsten, welcome. >> Hello Dave, great to be here with you today. >> Let's dive right in. What technologies and practices should we be thinking about that can help improve the security posture within the software supply chain? >> So I think the most important thing for folks to think about really is adopting DevSecOps. And while organizations talk about DevSecOps, and many folks have adopted DevOps, they tend to forget the security part of DevSecOps. And so for me, DevSecOps is both DevSec, how do I shift security left into my supply chain, and SecOps which is a better understood and more common piece of the puzzle, but then closing that loop between what issues are discovered in production and feeding that back to the development team to ensure that we're really addressing that supply chain. >> Yeah I heard a stat. I don't know what the source is, I don't know if it's true, but it probably is that around 50% of the organizations in North America, don't even have a SecOps team. Now of course that probably includes a lot of smaller organizations, but the SecOps team, they're not doing DevSecOps, but so what are organizations doing for supply chain security today? >> Yeah, I think the most common practice, that people have adopted is vulnerability scanning. And so they will do that as part of their development process. They might do it at one particular point, they might do it at more than one point. But one of the challenges that, we see first of all, is that, that's the only security gate that they've integrated into their supply chain, into their pipeline. So they may be scanning code that they get externally, they may be scanning their own code. But the second challenge is that the results take so much work to triage. This is static vulnerability scanning. You get information that is not in full context, because you don't know whether a vulnerability is truly exploitable, unless you know how exposed that particular part of the code is to the internet, for example, or to other aspects. And so it's just a real challenge for organizations, who are only looking at static vulnerability data, to figure out what the right steps to take are to manage those. And there's no way we're going to wind up with zero vulnerabilities, in the code that we're all working with today. Things just move too quickly. >> Is that idea of vulnerability scanning, is it almost like sampling where you may or may not find the weakest link? >> I would say that it's more comprehensive than that. The vulnerability scanners that are available, are generally pretty strong, but they are, again, if it's a static environment, a lot of them rely on NVD database, which typically it's going to give you the worst case scenario, and by nature can't account for things like, was the software that you're scanning built with controls, mitigations built in. It's just going to tell you, this is the package, and this is the known vulnerabilities associated with that package. It's not going to tell you whether there were compiler time flags, that may be mitigated that vulnerability. And so it's almost overwhelming for organizations, to prioritize that information, and really understand it in context. And so when I think about the closed loop feedback, you really want not just that static scan, but also analysis that takes into account, the configuration of the application, and the runtime environment and any mitigations that might be present there. >> I see, thank you for that. So, given that this digital risk and software supply chains are now front and center, we read about them all the time now, how do you think organizations are responding? What's the future of software supply chain going to look like? >> That's a great one. So I think organizations are scrambling. We've certainly at Red Hat, We've seen an increase in questions, about Red Hat's own supply chain security, and we've got lots of information that we can share and make available. But I think also we're starting to see, this strong increased interest, in security bill of materials. So I actually started working with, automation and standards around security bill of materials, a number of years ago. I participated in The Linux Foundation, SPDX project. There are other projects like CycloneDX. But I think all organizations are going to need to, those of us who deliver software, we're going to need to provide S-bombs and consumers of our software should be looking for S-bombs, to help them understand, to build transparency across the projects. And to facilitate that automation, you can leverage the data, in a software package list, to get a quick view of vulnerabilities. Again, you don't have that runtime context yet, but it saves you that step, perhaps of having to do the initial scanning. And then there are additional things that folks are looking at. Attested pipelines is going to be key, for building your custom software. As you pull the code in and your developers build their solutions, their applications, being able to vet the steps in your pipeline, and attest that nothing has happened in that pipeline, is really going to be key. >> So the software bill of materials is going to give you, a granular picture of your software, and then what the chain of, providence if you will or? >> Well, an S-bomb depending on the format, an S-bomb absolutely can provide a chain of providence. But another thing when we think about it, from the security angles, so there's the providence, where did this come from? Who provided it to me? But also with that bill of materials, that list of packages, you can leverage tooling, that will give you information about vulnerability information about those packages. At Red Hat we don't think that vulnerability info should be included in the S-bomb, because vulnerability data changes everyday. But, it saves you a step potentially. Then you don't necessarily have to be so concerned about doing the scan, you can pull data about known vulnerabilities for those packages without a scan. Similarly the attestation in the pipeline, that's about things like ensuring that, the code that you pull into your pipeline is signed. Signatures are in many ways of more important piece for defining providence and getting trust. >> Got it. So I was talking to Asiso the other day, and was asking her okay, what are your main challenges, kind of the standard analyst questions, if you will. She said look, I got great people, but I just don't have enough depth of talent, to handle, the challenges I'm always sort of playing catch up. That leads one to the conclusion, okay, automation is potentially an answer to address that problem, but the same time, people have said to me, sometimes we put too much faith in automation. some say okay, hey Kirsten help me square the circle. I want to automate because I lack the talent, but it's not, it's not sufficient. What are your thoughts on automation? >> So I think in the world we're in today, especially with cloud native applications, you can't manage without automation, because things are moving too quickly. So I think the way that you assess whether automation is meeting your goals becomes critical. And so looking for external guidance, such as the NIST's Secure Software Development Framework, that can help. But again, when we come back, I think, look for an opinionated position from the vendors, from the folks you're working with, from your advisors, on what are the appropriate set of gates. And we've talked about vulnerability scanning, but analyzing the configed data for your apps it's just as important. And so I think we have to work together as an industry, to figure out what are the key security gates, how do we audit the automation, so that I can validate that automation and be comfortable, that it is actually meeting the needs. But I don't see how we move forward without automation. >> Excellent. Thank you. We were forced into digital, without a lot of thought. Some folks, it's a spectrum, some organizations are better shape than others, but many had to just dive right in without a lot of strategy. And now people have sat back and said, okay, let's be more planful, more thoughtful. So as you, and then of course, you've got, the supply chain hacks, et cetera. How do you think the whole narrative and the strategy is going to change? How should it change the way in which we create, maintain, consume softwares as both organizations and individuals? >> Yeah. So again, I think there's going to be, and there's already, need request for more transparency, from software vendors. This is a place where S-bombs play a role, but there's also a lot of conversation out there about zero trust. So what does that mean in, you have to have a relationship with your vendor, that provides transparency, so that you can assess the level of trust. You also have to, in your organization, determine to your point earlier about people with skills and automation. How do you trust, but verify? This is not just with your vendor, but also with your internal supply chain. So trust and verify remains key. That's been a concept that's been around for a while. Cloud native doesn't change that, but it may change the tools that we use. And we may also decide what are our trust boundaries. Are they where are we comfortable trusting? Where do we think that zero trust is more applicable place, a more applicable frame to apply? But I do think back to the automation piece, and again, it is hard for everybody to keep up. I think we have to break down silos, we have to ensure that teams are talking across those silos, so that we can leverage each other's skills. And we need to think about managing everything as code. What I like about the everything is code including security, is it does create auditability in new ways. If you're managing your infrastructure, and get Ops like approach your security policies, with a get Ops like approach, it provides visibility and auditability, and it enables your dev team to participate in new ways. >> So when you're talking about zero trust I think, okay, I can't trust users, I got to trust the verified users, machines, employees, my software, my partners. >> Yap >> Every possible connection point. >> Absolutely. And this is where both attestation and identity become key. So being able to, I mean, the SolarWinds team has done a really interesting set of things with their supply chain, after they were, in response to the hack they were dealing with. They're now using Tekton CD chains, to ensure that they have, attested every step in their supply chain process, and that they can replicate that with automation. So they're doing a combination of, yep. We've got humans who need to interact with the chain, and then we can validate every step in that chain. And then workload identity, is a key thing for us to think about too. So how do we assert identity for the workloads that are being deployed to the cloud and verify whether that's with SPIFFE SPIRE, or related projects verify, that the workload is the one that we meant to deploy and also runtime behavioral analysis. I know we've been talking about supply chain, but again, I think we have to do this closed loop. You can't just think about shifting security left. And I know you mentioned earlier, a lot of teams don't have SecOps, but there are solutions available, that help assess the behavior and runtime, and that information can be fed back to the app dev team, to help them adjust and verify and validate. Where do I need to tighten my security? >> Am glad you brought up the SolarWinds to Kirsten what they're doing. And as I remember after 911, everyone was afraid to fly, but it was probably the safest time in history to fly. And so same analogy here. SolarWinds probably has learned more about this and its reputation took a huge hit. But if you had to compare, what SolarWinds has learned and applied, at the speed at which they've done it with maybe, some other software suppliers, you might find that they've actually done a better job. It's just, unfortunately, that something hit that we never saw before. To me it was Stuxnet, like we'd never seen anything like this before, and then boom, we've entered a whole new era. I'll give you the last word Kirsten. >> No just to agree with you. And I think, again, as an industry, it's pushed us all to think harder and more carefully about where do we need to improve? What tools do we need to build to help ourselves? Again, S-bombs have been around, for a good 10 years or so, but they are enjoying a resurgence of importance signing, image signing, manifest signing. That's been around for ages, but we haven't made it easy to integrate that into the supply chain, and that's work that's happening today. Similarly that attestation of a supply chain, of a pipeline that's happening. So I think as a industry, we've all recognized, that we need to step up, and there's a lot of creative energy going into improving in this space. >> Excellent Kirsten Newcomer, thanks so much for your perspectives. Excellent conversation. >> My pleasure, thanks so much. >> You're welcome. And you're watching theCUBE, the leader in tech coverage. (soft music)

(upbeat music) >> Hello everyone, my name is Dave Vellante, and we're digging into the many facets of the software supply chain and how to better manage digital risk. I'd like to introduce Kirsten Newcomer, who is the Director of Cloud and DevSecOps Strategy at Red Hat. Hello Kirsten, welcome. >> Hello Dave, great to be here with you today. >> Let's dive right in. What technologies and practices should we be thinking about that can help improve the security posture within the software supply chain? >> So I think the most important thing for folks to think about really is adopting DevSecOps. And while organizations talk about DevSecOps, and many folks have adopted DevOps, they tend to forget the security part of DevSecOps. And so for me, DevSecOps is both DevSec, how do I shift security left into my supply chain, and SecOps which is a better understood and more common piece of the puzzle, but then closing that loop between what issues are discovered in production and feeding that back to the development team to ensure that we're really addressing that supply chain. >> Yeah I heard a stat. I don't know what the source is, I don't know if it's true, but it probably is that around 50% of the organizations in North America, don't even have a SecOps team. Now of course that probably includes a lot of smaller organizations, but the SecOps team, they're not doing DevSecOps, but so what are organizations doing for supply chain security today? >> Yeah, I think the most common practice, that people have adopted is vulnerability scanning. And so they will do that as part of their development process. They might do it at one particular point, they might do it at more than one point. But one of the challenges that, we see first of all, is that, that's the only security gate that they've integrated into their supply chain, into their pipeline. So they may be scanning code that they get externally, they may be scanning their own code. But the second challenge is that the results take so much work to triage. This is static vulnerability scanning. You get information that is not in full context, because you don't know whether a vulnerability is truly exploitable, unless you know how exposed that particular part of the code is to the internet, for example, or to other aspects. And so it's just a real challenge for organizations, who are only looking at static vulnerability data, to figure out what the right steps to take are to manage those. And there's no way we're going to wind up with zero vulnerabilities, in the code that we're all working with today. Things just move too quickly. >> Is that idea of vulnerability scanning, is it almost like sampling where you may or may not find the weakest link? >> I would say that it's more comprehensive than that. The vulnerability scanners that are available, are generally pretty strong, but they are, again, if it's a static environment, a lot of them rely on NVD database, which typically it's going to give you the worst case scenario, and by nature can't account for things like, was the software that you're scanning built with controls, mitigations built in. It's just going to tell you, this is the package, and this is the known vulnerabilities associated with that package. It's not going to tell you whether there were compiler time flags, that may be mitigated that vulnerability. And so it's almost overwhelming for organizations, to prioritize that information, and really understand it in context. And so when I think about the closed loop feedback, you really want not just that static scan, but also analysis that takes into account, the configuration of the application, and the runtime environment and any mitigations that might be present there. >> I see, thank you for that. So, given that this digital risk and software supply chains are now front and center, we read about them all the time now, how do you think organizations are responding? What's the future of software supply chain going to look like? >> That's a great one. So I think organizations are scrambling. We've certainly at Red Hat, We've seen an increase in questions, about Red Hat's own supply chain security, and we've got lots of information that we can share and make available. But I think also we're starting to see, this strong increased interest, in security bill of materials. So I actually started working with, automation and standards around security bill of materials, a number of years ago. I participated in The Linux Foundation, SPDX project. There are other projects like CycloneDX. But I think all organizations are going to need to, those of us who deliver software, we're going to need to provide S-bombs and consumers of our software should be looking for S-bombs, to help them understand, to build transparency across the projects. And to facilitate that automation, you can leverage the data, in a software package list, to get a quick view of vulnerabilities. Again, you don't have that runtime context yet, but it saves you that step, perhaps of having to do the initial scanning. And then there are additional things that folks are looking at. Attested pipelines is going to be key, for building your custom software. As you pull the code in and your developers build their solutions, their applications, being able to vet the steps in your pipeline, and attest that nothing has happened in that pipeline, is really going to be key. >> So the software bill of materials is going to give you, a granular picture of your software, and then what the chain of, providence if you will or? >> Well, an S-bomb depending on the format, an S-bomb absolutely can provide a chain of providence. But another thing when we think about it, from the security angles, so there's the providence, where did this come from? Who provided it to me? But also with that bill of materials, that list of packages, you can leverage tooling, that will give you information about vulnerability information about those packages. At Red Hat we don't think that vulnerability info should be included in the S-bomb, because vulnerability data changes everyday. But, it saves you a step potentially. Then you don't necessarily have to be so concerned about doing the scan, you can pull data about known vulnerabilities for those packages without a scan. Similarly the attestation in the pipeline, that's about things like ensuring that, the code that you pull into your pipeline is signed. Signatures are in many ways of more important piece for defining providence and getting trust. >> Got it. So I was talking to Asiso the other day, and was asking her okay, what are your main challenges, kind of the standard analyst questions, if you will. She said look, I got great people, but I just don't have enough depth of talent, to handle, the challenges I'm always sort of playing catch up. That leads one to the conclusion, okay, automation is potentially an answer to address that problem, but the same time, people have said to me, sometimes we put too much faith in automation. some say okay, hey Kirsten help me square the circle. I want to automate because I lack the talent, but it's not, it's not sufficient. What are your thoughts on automation? >> So I think in the world we're in today, especially with cloud native applications, you can't manage without automation, because things are moving too quickly. So I think the way that you assess whether automation is meeting your goals becomes critical. And so looking for external guidance, such as the NIST's Secure Software Development Framework, that can help. But again, when we come back, I think, look for an opinionated position from the vendors, from the folks you're working with, from your advisors, on what are the appropriate set of gates. And we've talked about vulnerability scanning, but analyzing the configed data for your apps it's just as important. And so I think we have to work together as an industry, to figure out what are the key security gates, how do we audit the automation, so that I can validate that automation and be comfortable, that it is actually meeting the needs. But I don't see how we move forward without automation. >> Excellent. Thank you. We were forced into digital, without a lot of thought. Some folks, it's a spectrum, some organizations are better shape than others, but many had to just dive right in without a lot of strategy. And now people have sat back and said, okay, let's be more planful, more thoughtful. So as you, and then of course, you've got, the supply chain hacks, et cetera. How do you think the whole narrative and the strategy is going to change? How should it change the way in which we create, maintain, consume softwares as both organizations and individuals? >> Yeah. So again, I think there's going to be, and there's already, need request for more transparency, from software vendors. This is a place where S-bombs play a role, but there's also a lot of conversation out there about zero trust. So what does that mean in, you have to have a relationship with your vendor, that provides transparency, so that you can assess the level of trust. You also have to, in your organization, determine to your point earlier about people with skills and automation. How do you trust, but verify? This is not just with your vendor, but also with your internal supply chain. So trust and verify remains key. That's been a concept that's been around for a while. Cloud native doesn't change that, but it may change the tools that we use. And we may also decide what are our trust boundaries. Are they where are we comfortable trusting? Where do we think that zero trust is more applicable place, a more applicable frame to apply? But I do think back to the automation piece, and again, it is hard for everybody to keep up. I think we have to break down silos, we have to ensure that teams are talking across those silos, so that we can leverage each other's skills. And we need to think about managing everything as code. What I like about the everything is code including security, is it does create auditability in new ways. If you're managing your infrastructure, and get Ops like approach your security policies, with a get Ops like approach, it provides visibility and auditability, and it enables your dev team to participate in new ways. >> So when you're talking about zero trust I think, okay, I can't trust users, I got to trust the verified users, machines, employees, my software, my partners. >> Yap >> Every possible connection point. >> Absolutely. And this is where both attestation and identity become key. So being able to, I mean, the SolarWinds team has done a really interesting set of things with their supply chain, after they were, in response to the hack they were dealing with. They're now using Tekton CD chains, to ensure that they have, attested every step in their supply chain process, and that they can replicate that with automation. So they're doing a combination of, yep. We've got humans who need to interact with the chain, and then we can validate every step in that chain. And then workload identity, is a key thing for us to think about too. So how do we assert identity for the workloads that are being deployed to the cloud and verify whether that's with SPIFFE SPIRE, or related projects verify, that the workload is the one that we meant to deploy and also runtime behavioral analysis. I know we've been talking about supply chain, but again, I think we have to do this closed loop. You can't just think about shifting security left. And I know you mentioned earlier, a lot of teams don't have SecOps, but there are solutions available, that help assess the behavior and runtime, and that information can be fed back to the app dev team, to help them adjust and verify and validate. Where do I need to tighten my security? >> Am glad you brought up the SolarWinds to Kirsten what they're doing. And as I remember after 911, everyone was afraid to fly, but it was probably the safest time in history to fly. And so same analogy here. SolarWinds probably has learned more about this and its reputation took a huge hit. But if you had to compare, what SolarWinds has learned and applied, at the speed at which they've done it with maybe, some other software suppliers, you might find that they've actually done a better job. It's just, unfortunately, that something hit that we never saw before. To me it was Stuxnet, like we'd never seen anything like this before, and then boom, we've entered a whole new era. I'll give you the last word Kirsten. >> No just to agree with you. And I think, again, as an industry, it's pushed us all to think harder and more carefully about where do we need to improve? What tools do we need to build to help ourselves? Again, S-bombs have been around, for a good 10 years or so, but they are enjoying a resurgence of importance signing, image signing, manifest signing. That's been around for ages, but we haven't made it easy to integrate that into the supply chain, and that's work that's happening today. Similarly that attestation of a supply chain, of a pipeline that's happening. So I think as a industry, we've all recognized, that we need to step up, and there's a lot of creative energy going into improving in this space. >> Excellent Kirsten Newcomer, thanks so much for your perspectives. Excellent conversation. >> My pleasure, thanks so much. >> You're welcome. And you're watching theCUBE, the leader in tech coverage. (soft music)