 From around the globe, it's theCUBE with digital coverage of DevOps Virtual Forum, brought to you by Broadcom. Continuing our conversations here at Broadcom's DevOps Virtual Forum, Lisa Martin here, please welcome back to the program, Serge Luccio, the general manager of the Enterprise Software Division at Broadcom. Hey, Serge, welcome. Thank you, good to be here. So I know you were just participating with the BizOps manifesto that just happened recently. I just had the chance to talk with Jeffrey Hammond and he unlocked this really interesting concept that I wanted to get your thoughts on. Spiritual co-location as really a necessity for BizOps to succeed in this unusual time in which we're living. What are your thoughts on spiritual co-location in terms of cultural change versus adoption of technologies? Yeah, it's quite interesting, right? When we think about the major impediments for DevOps implementation that we talk about culture, right? And for the last 20 years, we've been talking about silos. We've been talking about the imperatives for these teams to put it with the line. In many ways, it's also much about these teams aligning, but about being in the same car in the same boat, right? It's really about using those teams around kind of the common purpose, the common objective. So to me, this is really about kind of changing this culture where people start to look at kind of OKRs as kind of the key objective that drives kind of the entire team. Now, what it means in practice is really that we need to change a lot of behaviors, right? It's not about your key, it's not about roles, it's about who can do what and when and driving advice towards action. It also means that we need, I mean, especially in these COVID times, it becomes very difficult, right? To drive kind of the kind of collaboration and affinity between these teams. And so I think there's a significant role that especially pools can play in terms of providing kind of discounted feedback for the teams to be in-depth but at a spiritual co-location. Well, and it talked about culture being, it's something that, you know, we're so used to talking about DevOps with respect to velocity, all about speed here. But of course, this time, everything changed so quickly but going from the physical spaces to everybody being remote really does take, it's very different, you can't replicate it digitally but there are collaboration tools that can kind of really be essential to help that cultural shift, right? Yeah, so to me, we tend to talk about collaboration in kind of a very mundane way, right? Of course, we can use Zoom, we can all get into the same room. But the point when, I think, when Jeff says spiritual co-location, it's really about do we all share the same objective? Do we, do we have a means for instance, our pipeline, right? When you talk about DevOps probably, we all start to think about kind of this continuous delivery pipeline that basically drives the automation and the orchestration across the team. But just think about the pipeline, right? At the end of the day, it's all about what is the meantime to feedback to these teams? If I'm a developer and I commit code, how long does it take for that code to be processed through the pipeline? How quickly can I get feedback? If I am a finance person, responding a product or a project, what is my meantime to feedback? And so a lot of kind of a, when we think about the pipeline, I think what's been really inspiring to me in the last year or so is that there's much more of an adoption of kind of door-to-back metrics. There is way more of a focus around downstream management. And to me, this is really when we talk about collaboration, it's really about how do you provide the feedback to the different stakeholders across the life cycle in a very timely manner. And that's really what we need to get to in terms of kind of this notion of collaboration. It's not too much about people being in the same kind of physical space. It's about, you know, when I check in code, you know, do I get the system to automatically identify what I'm going to break? If I'm about to release some application, how can the system help me reduce my change pillar rates because it's able to predict that some issue was introduced in the application or the product. So I think there's a great role of technology and AI can relate to actually provide kind of that new level of collaboration. So we'll get to AI in a second, but I'm curious what are some of the metrics? Do you think that really matter right now as organizations are still in some form, probably of transformation to this new, almost 100% remote workforce? So I'll just say first, I'm not a big fan of metrics. And the reason being that, you know, you can look at a change failure rate, right? Or a leak time or a cycle time. And those are interesting metrics, right? The trend on metrics is absolutely critical. But what's more important is, how do you get to the root cause? What is contributing to that metric to degrade or improve over time? And so I'm much more interested and we, you know, through Broadcom are way more interested in understanding what are the patterns that contribute to this? So I'll give a very mundane example. You know, we know that cycle time is heavily influenced by organizational boundaries. So, you know, we talk a lot about silos, but we've worked with many of our customers doing downstream mapping. And oftentimes what you see is that's really the boundaries of the organization that creates a lot of idle time, right? So to me, it's less about the metrics. I think the door metrics are a pretty, you know, valid set of metrics. But what's way more important is to understand what are the anti-parents? What are the things that we can detect through the data that actually are affecting those metrics? And I mean, over the last 10, 20 years, we've learned a lot about kind of what are the anti-parents within our large enterprise customers? And there are plenty of them. What are some of the things that you're seeing now with respect to patterns that have developed over the last seven to eight months? So I think the two areas which clearly are evolving very quickly are on kind of the front end of a life cycle where DevOps is more and more embracing value stream management, value stream mapping. And I think what's interesting is that in many ways the product is becoming the new silo. The notion of a product is very difficult by itself to actually define. People are starting to recognize that a value stream is not its own little kind of island. That in reality, when I define a product, this product oftentimes has dependencies on other products. And that in fact, you're looking at kind of a network of value streams, if you will. So on that end, there's clearly kind of a new set if you will have anti-patterns where products are being defined as a set of OKRs. They have interdependencies and you have a new set of silos. On the other hand, the other kind of key movement is around the SRE space where I think there is a cultural clash while the DevOps side is very much embracing this notion of OKRs and value stream mapping and value stream management. On the other hand, you have IT operations teams who still think business services, right? For them, they think about configure items, they think about infrastructure. And so, it's not uncommon to see teams where the operations team is still thinking about hundreds of thousands, tens of thousands of business services. And so there is kind of this boundary where I think, well, SRE is being put in place and there's lots of thinking about what kind of metrics can be defined. I think going back to culture, I think there is a lot of cultural evolution that's still required for true operations teams. And that's a hard thing. Cultural transformation in any industry, pandemic or not, is a challenging thing. You talked about AI and automation a minute ago. How do you think those technologies can be leveraged by DevOps leaders to influence their successes and their ability to collaborate and maybe see eye-to-eye with the SREs? Yeah, so there are kind of two, so even for myself as a leader of a 1500 people organization, there's a number of things I don't see on a daily basis. And I think the technologies that we have at our disposal today from the AI are able to mine a lot of data and expose a lot of issues that as leaders we may not be aware of. And some of these are pretty kind of easy to understand. We all think we're agile. And yet when you start to understand, for instance, what is the work in progress during the sprint? When you start to analyze the data, you can detect for instance that maybe the teams are over committed that there is too much work in progress. You can start to identify kind of inter-repancies either from a technology or from a people point of view which were hidden. And you can start to understand that maybe the changeable rate is dragging. So I believe that there is a fundamental role to be played by the tools to expose again, these antiparrots, to make these things kind of visible to the teams, to be able to even compare teams, right? One of the things that's amazing is now we have access to tons of data not just from a given customer but across a large number of customers. And so we start to compare kind of all of these teams kind of operate and what's working, what's not working. Thoughts on AI and automation as a facilitator of spiritual co-location? Yeah, absolutely. Absolutely, it's, you know, there's a, the problem we all face is the unknown, right? The velocity, the volume, the variety of the data, every day we don't really necessarily completely appreciate what is the impact of our actions, right? And so AI can really act as a safety net that enables us to understand what is the impact of our actions. And so yeah, in many ways, the ability to be informed in a timely manner to be able to interact with people on the basis of data and then collaborate on the data in the actual matter, I think it's a very powerful enabler in that respect. I mean, I've seen countless of times that for instance at the SRE boundary to basically show that what's one of the quality attributes of an incoming release, right? And exposing that to an operations person and an SRE person and enabling that collaboration and dialogue through the data is a very, very powerful tool. Given your recommendations for how teams can use, the SRE folks, the DevOps says, can use AI and automation in the right ways to be successful rather than some ways that are going to be non-productive? Yeah, so to me, there's a part, the question really is when we talk about data, there are different ways you can use data, right? So you can do a lot of analytics, predictive analytics. So I think there is a tendency to look at, let's say a specific KPI, like an availability KPI or change filler rate and to basically do a rational analysis and project, all these things going to happen in the future. To me, that's a bad approach. The reason why I fundamentally think it's a bad approach is because our systems, the way we develop software is a non-linear kind of system, right? Software development is not linear in nature. And so I think this is probably the worst approach, is to actually focus on metrics. On the other hand, if you start to actually understand at a more granular level, what are the things which are contributing to this, right? So if you start to understand, for instance, that whenever maybe you affect a specific part of the application, that translates into production issues. So I've actually a customer who identified that over 50% of their unplanned outages were related to specific components in their architecture. And whenever these components were changed, these resulted in these unplanned outages. So if you start to be able to basically establish causality, right, cause and effect between kind of data across the lifecycle, I think this is the right way to use AI. And so fundamentally, I think it's way more about kind of a classification problem. What are the classes of problems that do exist and affect things as opposed to analytics predictive, which I don't think is as powerful? So I mentioned in the beginning of our conversation just came off the biz-ops manifesto. You're one of the authors of that. I want to get your thoughts on dev-ops and biz-ops overlapping, complimenting each other. What, from the biz-ops perspective, what does it mean to the future of dev-ops? Yeah, so it's interesting, right? So if you think about dev-ops, there's no founding documents, right? I mean, we can refer to the Phoenix project. I mean, there are a set of documents which have been written, but in many ways there's no clear definition what dev-ops is. If you go to the dev-ops institute today, you will see that they have specific trainings, for instance, on value-shoot management, on SRE. And so in many ways, the problem we have as an industry is that there are set of practices between agile, dev-ops, SRE, value-shoot management, ITAL, right? And we all basically talk about the same things, right? We all talk about, essentially, accelerating the meantime to feedback, but yet we don't have the common framework to talk about that. The other key thing is that we had to wait for Gene's last book to really start to get into the business aspect, right? And for value-shoot mapping to start to emerge for us to start as an industry, right? IT to start to think about, what is our connection with the business aspect? What's our purpose, right? And ultimately it's all about driving these business outcomes. And so to me, dev-ops is really about putting a lens on this critical element, that it's not business and IT that we in fact need to fuse business and IT, that I need bits to transform itself to recognize that it's this value generator, right? It's not a cost center. And so the relationship to me, it's more that business provides kind of this overall kind of framework, if you will, that set the context for what is the reason for IT to exist, what are the core values and principles that IT needs to embrace to again, change from cost center to a value center. And then we need to start to use this as a way to start to unify some of, again, the core practices, whether it's agile dev-ops, value-shoot mapping, SRE. So I think over time, my hope is that we start to harmonize a lot of our practices, language and cultural elements. Last question, Serge, in the last few seconds we have here, talking about the relation between dev-ops and dev-ops, what do you think as dev-ops evolves and as you talk just through some of your insights, what should our audience keep their eyes on in the next six to 12 months? So to me, the key challenge for the industry is really around, so we're seeing a very rapid shift towards kind of project to product, right? What we don't want to do is to recreate these new silos, these hard silos. So that's one of the big changes that I think we need to be really careful about because ultimately it is about culture. It's not about how we segment the work, right? And it's true culture that we can overcome silos. So back to, I guess, with Jeffrey's concept of the spiritual co-location, I think it's really about that. It's really about focusing on the business outcomes, on aligning, on driving engagement across the teams, but not to create kind of a new set of silos, which instead of being vertical are going to these horizontal products. Created by Serge that looking at culture as kind of a way of really addressing and helping to reduce, replace challenges. We thank you so much for sharing your insights and your time at today's dev-ops spiritual forum. Thank you. Thanks for your time. For Serge Lucio, Lisa Martin, we'll be right back.