 Live from Vancouver, Canada, it's theCUBE. Covering OpenStack Summit North America 2018. Brought to you by Red Hat, the OpenStack Foundation, and its ecosystem partners. Welcome back, I'm Stu Miniman with my co-host John Troyer and you're watching theCUBE's coverage of OpenStack Summit 2018 here in Vancouver. Happy to welcome to the program first time guest Roland Cabana who's a DevOps manager at Vault Systems out of Australia. But you come from a little bit more local. Thanks for joining us Roland. Thank you, thanks for having me. Yes, I'm actually born and raised in Vancouver, moved to Australia a couple years ago, realized the potential in Australian cloud providers and been there ever since. All right, so one of the big things we talk about here at OpenStack of course is, do people really build clouds with this stuff? Where does it fit? How's it doing? So nice lead into what does Vault Systems do if there are people that aren't aware? Definitely, so yes, we do build cloud, a cloud or many clouds actually. And Vault Systems provides cloud services, infrastructure service to Australian government. And we do that because we are a certified cloud. We are certified to handle unclassified DLM data and protected data. And what that means is the sense of information that's gathered for the Australian citizens and anything to do with big user space data is actually secured with certain controls set up by the Australian government. The Australian government body around this called ASD, the Australian Signals Directorate and they release a document called the ISM. And this document actually outlines a thousand and 88 plus controls that dictate how a cloud should operate, how data should be handled inside of Australia. Yeah, just to step back for a second. Yeah, I took a quick look at your website. It's not like you're listed as the government OpenStack cloud there. Can you give us, where does OpenStack fit into the overall discussion of the identity of the company? What your ultimate end users think about how they're doing? Help us kind of understand where this fits. Yeah, for sure. And I mean, the journey started long ago when we actually, our CEO, Rupert Taylor Price, set out to handle a lot of government information and tried to find this cloud provider to kind of handle it in the prescribed way that the Australian Signals Directorate needed to handle. So he went through different vendors, different cloud platforms, and found out that you couldn't actually meet all the controls in this document using a proprietary cloud, or using a proprietary platform deployed on your bare metal hardware. So eventually he found OpenStack and saw that there was a great opportunity to massage the code and change it so that it would comply 100% to the Australian Signals Directorate, ISM. All right, so the keynote this morning we were talking about people that build people, that operate, you've got DevOps in your title. Tell us a little bit about your role in working with OpenStack specifically and in broader scope of your... For sure, for sure. So in vault systems, the DevOps manager, and so what I do is we run through a lot of tests in terms of our infrastructure. So, complying to those controls, like I mentioned earlier, going through the rigmarole of making sure that all the different services that are provided on our platform comply to those specific standards, or those specific use cases. So as a DevOps manager, I handle a lot of the pipelining in terms of where the code goes. I handle a lot of the logistics and operations. And so it actually extends beyond just operation and development. It actually extends into our policies. And so marrying all that stuff together is pretty much my role day-to-day. I have a leg in the infrastructure team with the engineering, and I also have a leg with the solutions architects and how they get feedback from different customers in terms of what we need and how we architect that so it's safe and secure for government. Roland, since one of your parts of your remit is compliance, would you say that you're DevSecOps? Are you like that one or not? Well, I guess there's a few more buzzwords and some pretty more roles I can throw in there. But yeah, I guess yes, DevSecOps. There's a strong security posture that vault holds. And we hold it to a higher standard than a lot of the other incumbents or a lot of platform providers because we are actually very sensitive about how we handle this information for government. So security is a big portion of it. And I think the company culture internally is actually centered around how we handle the security. A good example of this is internally we actually have controls about printing. Most modern companies today, they print pages and it's an eco thing. It's an eco thing for us too, but at the same time, there are controls around printed documents and how sensitive those things are. And so our position in the company is if that control exists because the Australian government decides that that's a sensitive matter, let's adopt that in our entire internal ecosystem. There was a lot of talk this morning at the keynote, both about upgrades, and I'm blanking on the name of the new feature, but also about Zool and about upgrading OpenStack. You guys are a full upstream OpenStack expert cloud provider. How do you deal with upgrades? And what do you think the state of the OpenStack community is in terms of kind of upgrades and maintenance and day two kind of stuff? Well, I'll tell you the truth. The upgrade path for OpenStack is actually quite difficult, right? I mean, there's a lot of moving parts, a lot of components, and you have to be very specific in terms of how you upgrade to the next level. If you're not keeping in step with the next releases, you may fall behind and you can't upgrade Keystone from a Liberty all the way up to Ocata, right? You're basically stuck there. And so what we do is we try to figure out what the government needs. What are the features that are required? And it's also a conversation piece with government because we don't have certain features in this particular release of OpenStack. It doesn't mean we're not going to support, we're not going to move to the next version just because it's available, right? There's a lot of security involved in infusing our controls inside our distribution of OpenStack. I guess you can call it the distribution on our build of OpenStack. But it's all based on a conversation that we start with the government. So if they need VGPUs for some reason, right? With the Queens release that's coming out, that's a conversation we're starting and we will build into that functionality as we need. So does that mean that you have different instances with different versions? And if so, how do you manage all of that? Well, okay, so yes, that's true. We do have different versions with where we have a Liberty release or we have a Ocata release, which is predominant in our infrastructure. And that's only because we started with the inception of the Liberty release before our certification process. A lot of the things that we work with government for is how do they progress through this cloud maturity model? And the forklift and shift is actually a problem when you're talking about releases. But when you're talking about containerization, you're talking about agile methodologies and things like that, it's less of a reliance on the version because you now have the ability to respond that same application, migrate the data and have everything live as you progress through different cloud platforms. And so as OpenStack matures, this whole idea of the fast forward idea of getting to the next release, because now they have an integration step where they have a path to the next version, even though you're two or three versions behind. Because let's face it, most operators will not go to the latest and greatest because there's a lot of issues that you're going to face there. I mean, not that the software is bad, it's just that early adopters will come with early adopter problems. And you need that user base, you need those forum conversations to be able to be safe and secure about whether or not you can handle those kinds of things. And there's no need for our particular user space to have those latest and greatest things unless there's an actual request. Rolan, you are an IAS provider. How are you handling containers or requests for containers from your customers? Yes, containers is a big topic. There's a lot of maturity happening right now with government in terms of what a container is, for example, what is orchestration with containers? How does my legacy application for collecting shift to a container? And so we're handling in stages, right? Because we're working with government in their maturity. We don't do container services on the platform, but what we do is we open source a lot of code that allows people to deploy, let's say a Terraform file that creates a Docker host. And we give them examples, a good segue into what we've just launched last week was our Vault Academy, which we are now training 3,000 government public servants on new cloud technologies. We're not talking about how does an OS work. We're talking about infrastructures, code, we're talking about Kubernetes, we're talking about all these cool fun things all the way up to function as a service, right? And those kinds of capabilities is what's going to propel government in Australia moving forward in the future. Wow, that's great. You hit on one of my hot buttons here. So function as a service, do you have serverless deployed in your environment or is it an education at this point? It's an education at this point. Right now we have customers who would like to have that available as a native service in our cloud. But what we do is we concentrate on the controls and the infrastructure as a service platform first and foremost, just to make sure that's secure and it's compliant. Everyone has the ability to deploy functions of service on their accounts or on their tendencies and have that available to them through a different set of APIs. There's a whole bunch of open source versions out there. Is that what they're doing? Do you have any preference towards the open whisk or FN or, you know, fission or all the different versions that are out there? I guess you can sort of like, pick your racehorse in that regard because it's still early days. And I think open whisk is pretty much what I've been looking at recently and it's just a discovery stage at this point. There are more mature customers who are coming in, some partners who are, you know, championing different technologies. So the great thing is that we can make sure that our platform is secure and they can build on top of it. All right. Yeah, just you brought up security again. One of the areas I wanted to poke at a little bit is your network. Yes. So being an IaaS provider, networking's critical. What are you doing from a networking standpoint? Is micro segmentation part of your environment? Definitely. So natively we build in our cloud, the functions that we build in our cloud are all around security, obviously. Micro segmentation is a big part of that. Training people in terms of how micro segmentation works from a forklift and shift perspective. And the network connectivity we have with government is also part of this whole model, right? And so we use technologies like Melanox for 100G fabric. We're a BGP internally. So we're routing through the host or routing to the host. And we have this, so in Australia there's this, there's service from the Department of Finance. They created this idea of an icon network. And what it is, it's an actually direct media fiber from the department directly to us. And that makes directly to the edge of our cloud and pipes right through into their tenancy. So essentially what happens is this is true, true hybrid cloud. I'm not talking about going through gateways and stuff. I'm talking about, I spin up an instance in the vault cloud and I can ping it from my desktop in my agency. Low latency, sub millisecond, direct fiber link up to 100G. Yeah. Do you have certain programmability you're doing in your network? I know lots of service providers, they want to play and get in there. They're using, you know, new operating models. Yes. I mean, we're using the, I draw a blank. There's a lot of technologies we're using for network and the Cumulus Networking OS is what we're using. That allows us to bring it into our automation team and actually use more of a DevOps tool to sort of create the deployment from a code perspective instead of having a lot of engineers hard coding things right on the actual production systems. Which allows us to get a lot of the changes, which is part of the whole security posture as well. So we're doing a lot of network offloading on the ConnectX five cards in the data center. We're using Cumulus Networks for bridging. We're working with Neutron to make sure that we have Neutron routers and making sure that that's secure and it's code reviewed. And, you know, there's a lot of moving parts there as well. And I think, you know, from a security standpoint and from a network functionality standpoint, we've come to a happy place in terms of providing the fastest network possible and also the most secure and safe network as possible. Roland, you're working directly with the upstream OpenStack projects. And it sounds like some others as well. You're not working with a vendor who's packaging it for you or supporting it. Right. So that's a lot of responsibility on you and your team. Yes. So I'm curious how you work with the OpenStack community and how you've seen the OpenStack community develop over the years. Yes. I mean, we have a lot of talented people in our company who actually take OpenStack as a passion, right? This is what they do. This is what they love. They've come from different companies who worked in OpenStack and have contributed a lot actually to the community. And that actually that segues into how we operate inside culturally in our company is that if we do work with upstream code and it doesn't have anything to do with the security compliance of the Australian Signals Directorate in general, we'd like to upstream that as much as possible, which contribute back to code where it seems fit, right? Obviously, there's vendor mixes and things that we have internally. And that's with the melanogs and cumulus stuff, but anything else beyond that is usually contributed up. Our team's actually very supportive of each other. We have network specialists. We have storage specialists. And it's a culture of learning. So there's a lot of synchronizations, a lot of synergies inside the company. And I think that's part to do with the people who make up vault systems. And that whole camaraderie is actually propagated through our technology as well. One of the big themes of the show this year has been kind of broadening out of what's happening. We talked a little bit about containers already. Edge computing is a big topic here. Either Edge or some other areas, what are you looking for next from this ecosystem or new areas that Vault is looking at poking at? Well, I mean, a lot of exciting things for me personally, I guess. I can't talk to Vault in general, but because there's a lot of engineers who have their own opinions of what they like to see. But with the Queen's release with the VGPU is something I like that was great. A long-term release cycle with the OpenStack Foundation would be great. Or the OpenStack platform would be great. And that's just to keep in step with the next releases to make sure that we have the continuity even though we're missing one release. There's a jump point. Can you actually put a point on that? What that means for you? We talked to Mark Carlyer a little bit about this morning, but explain what you're looking and why that's important. Well, it comes down to user acceptance, right? So, I mean, let's say you have a new feature or a new project that's integrated into OpenStack. And some people find out that there's these new functions that are available. There's a lot of testing behind the scenes that has to happen before that can be vetted and exposed as part of our infrastructure as a service platform. And so, by the time you get to the point where you have all the checks and balances and marrying that next to the Australian controls that we have, it's one year, two years, or however it might be. And by that time, we're at another release. And so, you do all that work. You want to make sure that you're not doing that work and refactoring it for the next release when you're ready to go live. And so, having that long-term release is actually what I'm really keen about. Having that point of, that jump point to the latest and greatest. Yeah, well, Roland, I think that's a great point. It used to be we were on the 18-month cycle. OpenStack was more like a six-month cycle. So, absolutely understand why it's important that I don't want to be tied to a release when I want to get a new function. All right. That's right. Roland Cabana, thank you for giving us the insight into vault systems and congrats on all the progress you have made. So, for John Troyer, I'm Stu Benjamin. Back here with lots more coverage from the OpenStack Summit 2018 in Vancouver. Thanks for watching theCUBE.