The guide to DevOps: “Teamwork is what really drives DevOps”
The Atlassian Summit is behind us but now’s the perfect time to roll out the interviews with Atlassian executives. We talked to Roger Barnes, Group Product Manager of DevOps & ITSM Solutions at Atlassian about everything DevOps, including pillars such as monitoring and automation, the importance of Ops and more.
JAXenter: Could you tell us more about your job at Atlassian?
Roger Barnes: I’m a group product manager on the server side of the business, so I’m responsible for how our products operate today and in the future across DevOps and IT. I focus on the use cases as much as the individual products.
JAXenter: Why should companies switch to DevOps if everything is presumably working fine?
Roger Barnes: I’d like to see if that was the case! If everything is working fine, then I would suggest that delivering speed and quality should be optimized as can be. In which case, they probably are already doing DevOps. It’s a mindset, so it’s likely they are already thinking that way. It’s not exactly a new concept. Looking back twenty years, there are people who were delivering things to production every day in certain environments with elements of that quality, speed, and automation. It was done with less sophisticated tooling though.
So, there’s a likelihood they are doing DevOps already. Ultimately, these things are always a journey. If you think you’ve reached the top of the mountain, chances are you are not looking hard enough at how much further you could go.
Quite often the things we do for the first time we will do in the cloud because we can learn how to integrate really fast and then we can bring the things that work to the server and data platforms.
JAXenter: Let’s talk about Bitbucket. What are the top things that new developers should know about Bitbucket before they start using it?
Roger Barnes: Fundamentally, it’s built upon Git. Git itself is great for facilitating collaborative and distributed development. What Bitbucket builds on top of that is a collaborative workflow.
The second thing is that you can now have a more team-based environment where you have people working in branches, doing their code review and really collaborating and following a process where code gets reviewed the right amount at the right time. It’s done in a way that helps move things forward as quickly as possible.
JAXenter: What does Bitbucket have that its competitors -for example, GitHub or GitLab- don’t?
Roger Barnes: I can’t speak to what others don’t have, but I do think that some of the strengths of Bitbucket are the server and offerings of our cloud as well. We do have a very strong focus on stability, performance, and scale.
Then there are things like the integration of our tools. We have not just in Bitbucket a strong collaboration tool, but there are things like Jira and Bamboo. When you connect these things together, you really get the power of those individual tools being more than the sum of the individual parts. This means that developers get a more complete workflow and it becomes more of a collaboration platform.
On top of that, I think that some of the ways that we approach integration are quite unique in that people can really extend their products in quite powerful ways and in a DevOps world that’s really important because each team has unique needs on top of the foundation of collaboration. If you’re doing mobile apps or web development you will need to plug in different things. So, now integration capabilities there are solid.
We have a DevOps marketplace with over 200 integrations and add-ons. If there’s something there that you don’t see and you need, you can also go ahead and write your own as well.
JAXenter: Let’s combine the two: DevOps and Bitbucket. How can Bitbucket build a DevOps culture?
Roger Barnes: I think it can help facilitate it. Obviously, a lot of culture comes from the people and their mindset in the first place. Bitbucket certainly encourages it along those lines in that it provides the necessary collaboration. Work in Bitbucket mainly revolves around a pull request. If you take the idea of agile, delivering small items of work, and reducing risk, Bitbucket’s goal is basically to make that possible when you get to the code stage. So you can do those small independent changes, you can have them collaboratively reviewed, you can control which bits you merge and when.
Basically, you can make sure that all that agile that you put together in planning is actually put into delivery and of course that connects to your CI. If you have all of those pieces working together, you actually get the benefit of all that agile effort.
JAXenter: There’s been a change to Bitbucket cloud. What’s the biggest benefit of this redesign? Why the focus on code reviews?
Roger Barnes: It’s the core of what Bitbucket needs to do really well. The cloud tiering in particular recognizes that to do some of the future things we want to do, we needed to rethink things a little bit and change how it works. Basically, we want to establish a platform for the future of the code review platform that we have.
In addition, when dealing with pull review, there’s a lot that you are working with at once. So, a lot of the thinking that’s gone into the new design is about putting the right things in the right places, making it easy to get around, and making it easier to feel confident that you’ve seen everything that you need to see. Ultimately, the team also saw an improvement in the time needed to approve. Now, the reviewer that’s coming into a pull request is able to approve it more quickly than they would before.
JAXenter: Are there any other announcements related to Bitbucket or is there something new coming soon?
Roger Barnes: One thing we talked about was Bitbucket server and data center; we are planning to release something called Code Insights. In the DevOps world, there’s this notion of shifting left and a lot of build systems, static analysis tools, security scanning tools, and others are down the pipeline analyzing changes and producing some sort of output about the health of the code. But that often happens further down the pipe where people have to go into those systems to look for that information.
With the Code Insights work that we have done for Bitbucket, you will be able to see that information in a pull request. So, at the same time that reviewers are reviewing each other’s code, the systems get a chance to say that there might be a vulnerability with a particular approach to code. This would help with catching it at the exact same time that the peer review would. This information has existed but it is usually somewhere else. Now it will be brought front and center. We are really excited about it!
“DevOps is a somewhat unfortunate title”
JAXenter: Circling back to DevOps, we cannot talk about DevOps and not mention containers and serverless. There are so many versions of the truth: is serverless the future of containers? Or are containers the future of DevOps? What is your version of the truth?
With serverless, you still have to write some code, there’s just less packaging around it. So that doesn’t mean you can’t use peer, you can still use code review.
Roger Barnes: This is something of a crystal ball gaze. I think serverless is seeing some benefits in certain areas, especially at the forefront of this evolution of things. Then you look at how the rest of the world follows and there’s still a place for containers of course. There are still people using virtual machines and more traditional virtualized infrastructure and then there are others that are just still running on bare metal for various reasons.
I think these things will find a lot of success for particular use cases and some of them will prove super successful. I think containerization is a very obvious new way to package up applications. But it’s not necessarily evident yet that serverless is going to replace it. It introduces a whole set of new challenges and ways of working that need to be resolved so we have to see where that goes.
This is already true in monolithic applications, but for any individual or team to make an assertion about a system, they need to trust the tools. These things are already getting bigger and more complex. The ability to make assertions on how it’s all going to work, when you’ve decomposed it and put it into little tiny pieces, is a very different way of producing software. That said, it’s really exciting.
We use different techniques for our products, especially in our cloud products. With serverless, you still have to write some code, there’s just less packaging around it. So that doesn’t mean you can’t use peer, you can still use code review. Continuous integration and testing becomes even more important and how you do that may change a little bit, but at its core that still needs to happen. We are very happy to adapt our tools to different use cases as they change and evolve.
JAXenter: Why does Atlassian focus so much on the cloud? Is the cloud just a passing trend or is it here to stay?
Roger Barnes: There is almost certainly going to be more people in the cloud in the future. We are investing to make sure that’s where people end up. There will always be people who have genuine reasons and will benefit from our server and data center offerings as well.
Not all of our focus is on the cloud, but the cloud is where the innovation tends to happen because we can learn more quickly there. So, quite often the things we do for the first time we will do in the cloud because we can learn how to integrate really fast and then we can bring the things that work to the server and data platforms. We just announce them less loudly!
JAXenter: Last year in an interview, Maciej Winnicki from Serverless Inc. said that “Serverless is a revolution of the cloud”. What is your take on that?
Roger Barnes: It’s hard to disagree because it’s part of the definition, I guess! It doesn’t necessarily exclude the idea that clouds don’t have to be public. They can be private. People are running all sorts of infrastructure in their data centers for all sorts of reasons, and there’s no reason that serverless can’t happen there as well. I think a lot of new tech product development, even for internal use cases, can start utilizing it as a result. It is still going to take some time.
JAXenter: Is there anything else that you would like to share?
Roger Barnes: The main thing that we want to get across to people is that teamwork is what really drives DevOps. Whether that’s culture practices or tools, ultimately it is about how multiple people work together. Given that’s our focus as a company, not just tools, I see that as a really exciting space to be playing in. Already today you can do a lot.
Our playbook, for example, helps teams understand their health. We can connect people’s tools together using our marketplace. There’s this foundation that helps people plan, track, deliver, and monitor. We think this is quite powerful and quite adaptable as well, which we need to be because this world is ever-chaotic and complex.
JAXenter: There’s a lot of emphasis on Ops at Atlassian. Usually, the focus is on developers, but more and more people are asking about Ops.
Roger Barnes: It’s interesting because DevOps is a somewhat unfortunate title. It is supposed to be a much broader mindset than the word would suggest. People are attacking the Dev parts, the Ops and parts, and each time we do we put pressure on the next adjacent part of the overall delivery. DevOps really wants to make sure this mindset is the case for everyone involved, from IT up to production and back again. This includes things like service, management, incidents, and operations that should all be embraced by the DevOps mindset.
It’s just a matter of time that we improve people’s understanding and tooling that goes along with it.
JAXenter: Are we creating silos when we first talk about developers and then about Ops? The focus is often on each one by one, instead of at the same time.
DevOps really wants to make sure this mindset is the case for everyone involved, from IT up to production and back again.
Roger Barnes: Possibly. Because it’s such a big thing, people can’t talk about all of it. Or if they do, they don’t talk about it very specifically. As soon as you get specific, you end up focusing on an area. When I got more into the DevOps mindset myself, in terms of what I had to think about, I still couldn’t help but think about my developer background. So, I always thought about Bitbucket and Bamboo.
Now I’m responsible for IT and the service desk is also part of that world. Of course, there is also the operations side of things. So yes, that’s right, there has been more attention on the development side. It’s probably because sometimes the technology drives the conversation a bit more. Things like configuration also made its own push into trying to establish itself as what DevOps is about. Software did it as well, when really it is about all of these things combined together.
JAXenter: How can we go from a tool-oriented attitude to a more team-oriented attitude?
Roger Barnes: It’s a challenge. It’s tricky, largely as a software company Atlassian makes tools. We also provide the playbook and we work with partners to provide guidance. But it all has to start with the right mindset. How do you hire and retain people? Or, how do you help people who are used to working a certain other way? What about helping people who don’t want to enter this mindset? You have to find something positive that they can do.
That’s a challenge and I don’t think there is as much energy put into that, probably because it’s harder to package up and give to customers and companies. Sometimes they have to work it out themselves because everyone is in a very different place. That means less specific advice, guidance, and thought leadership goes into it. Whereas tools you can say, assuming you have this mindset, here’s how the tools work.
Don’t miss our “Guide to DevOps” interview series with Atlassian executives: