The next BriefingsDirect analyst thought leadership discussion explores the pitfalls and payoffs of DevOps adoption — with an emphasis on the developer perspective.
We’re joined by two prominent IT industry analysts, the founders of RedMonk, to unpack the often tricky path to DevOps and to explore how enterprises can find ways to make pan-IT collaboration a rule, not an exception.
With that, please join me in welcoming James Governor, Founder and Principal Analyst at RedMonk, and he is based in London, and Stephen O’Grady, also Founder and Principal Analyst at RedMonk, and he is based in Portland, Maine. The discussion is moderated by me, Dana Gardner, Principal Analyst at Interarbor Solutions.
Here are some excerpts:
Gardner: Gentleman, let’s look at DevOps through a little bit of a different lens. Often, it’s thought of as a philosophy. It’s talked about as a way of improving speed and performance of applications and quality, but ultimately, this is a behavior and a culture discussion — and the behavior and culture of developers is an important part of making DevOps successful.
What do developers think of DevOps? Is this seen as a positive thing or a threat? Do they have a singular sense of it, or is it perhaps all over the map?
O’Grady: The overwhelming assessment from developers is positive, simply because — if you look at the tasks for a lot of developers today — it’s going to involve operational tasks.
In other words, if you’re working, for example, on public-cloud platforms, some degree of what you’re doing as a developer is operational in nature, and vice versa, once you get to the operational side. A lot of the operational side has now been automated in ways that look very much like what we used to expect from development.
So there is very naturally a convergence between development and operations that developers are embracing.
Driven by developers
Governor: I think developers have driven the change. We’ve seen this in a number of areas, whether it’s data management or databases, where the developers said, “We’re not going to wait for the DBA anymore. We’re going to do NoSQL. We’re just going to choose a different store. And we’re not going to just use Oracle.” We’ve seen this in different parts of IT.
The bottom line is that waterfall wasn’t working. It wasn’t leading to the results it should, and developers were taking some of the heat from that. So engineers and developers have begun to build out what has now becomes DevOps. A lot of them were cloud natives and thought they knew best, and in some cases, they actually did some really good work.
Partly enabled by cloud computing, DevOps had made a lot of sense, because you’re able to touch everything in a way that you weren’t able to on your own prem. It has been a developer-led phenomenon. It would be surprising if developers were feeling threatened by it.
Gardner: Enterprises, the traditional Global 2000 variety, see what happens at startups and recognize that they need to get on that same speed or agility, and oftentimes those startups are developer-centric and culturally driven by developers.
If the developers are, in fact, the tip of the arrow for DevOps, what is it that the operations people should keep in mind? What advice would you give to the operations side of the house for them to be better partners with their developer core?
Governor: The bottom line is that it’s coming. This is not an option. An organization could say we have this way of doing ops and we will continue doing that. That’s fine, but to your point about business disruption, we don’t have time to wait. We do need to be delivering more products to market faster, particularly in the digital sphere, and the threat posture and the opportunity posture have changed very dramatically in the past three years.
We’ve all heard that software is eating the world, but what that basically says is that the threats are real. We used to be in an environment where, if you were a bank, you just looked at your four peer banks and thought that as long as they don’t have too much of an advantage, we’re okay. Now they’re saying that we’re a bank and we’re competing with Google and Facebook.
Actually, the tolerance for stability is a little bit lower than it was. I had a very interesting conversation with a retailer recently. They had talked about the different goals that organizations have. And it was very interesting to me that he said that, on the first day they launched a new mobile app, it fell over. And they were all high-fiving and fist pumping, because it meant they had driven so much traffic that the app fell over, and it was something they needed to remediate.
That is not how IT normally thinks. Frankly, the business has not told IT they want it to be either, but it has sort of changed. I think the concern for new experiences, new digital products is now higher than the obsession with stability. So it is a different world. It is a cultural shift.
Gardner: Whether you’re a bank or you’re making farm equipment, your software is the biggest differentiator you can bring to the market. Stephen, any thoughts about what operations should keep in mind as they become more intertwined with the developer organization?
O‘Grady: The biggest thing for me is a variety of macro shifts in the market, things like the availability of open-source software and public cloud. It used to be that IT could control the developer population. In other words, they were essentially the arbiter of what went to production and what actually got produced. If you’re a developer and you have a good idea, but you don’t have any hardware or infrastructure, then you’re out of luck.
These days, that’s changed, and we see these organizationally, where developers can go to operations and say, they need infrastructure, and operations will say six months. The developers say, “To hell with six months. I’m going to go to Amazon and I have something up in 90 seconds.” The notion that’s most important for operations is that they’re going to have to work with developer populations because, one way or another, developers are going to get what they want.
Gardner: When we think about the supplier, the vendor, side of things, almost every vendor I’ve talked to in the last two or three months has raised the question of DevOps. It has become top of mind for them. Yet, if you were to ask an organization, how do you install DevOps, how do you buy DevOps, which shape box does it come in, none of those questions are relevant because it’s not a product.
How do the vendors grease the skids toward adoption, if you will? What do you think needs to happen from those tools, platforms and technologies?
Governor: It’s very easy to say that DevOps is not a product, and that’s true. On the other hand, there are some underlying technologies that would have driven this, particularly in automation and the notion of configuration is code.
One of the things to understand is that in the world of open source, it’s perhaps going to be packaged by a more traditional vendor. Certainly, one of the things is rethinking how you do automation. I would counsel anyone in IT ops to at least have a team starting to look at that, perhaps for some new development that you’re doing.
It’s easy to say that everything is a massive transformation, because then it’s just a big services opportunity and there’s lots of hand waving. But at the end of the day, DevOps has been a tools-driven phenomenon. It’s about being closer to the metal, having better observability, and having a better sense of how the application is working.
One of the key things is the change in responsibility. We’ve lived in an environment where we remember the blame game and lots of finger pointing. If you look at Netflix, that doesn’t happen. The developer who breaks the build fixes it.
There are some significant changes in culture, but there are some tools that organizations should be looking at.
What can they do?
O’Grady: If we’re talking from a vendor perspective, they can talk to their customers about the cultural change and organizational change that’s necessary to achieve the results that they want, but they can’t actually affect that. In other words, what we’re talking about, rather, is what they can do.
The most important thing that vendors who play in this and related spaces can do is understand that it’s a continuum. From the first code that gets written, to check-in, to build, to being deployed on to infrastructure that’s configured using automated software, it’s a very, very long chain, a very long sequence of events.
Understanding from a vendor perspective where you fit into that lifecycle, what are the other pieces you have to integrate with, and from a customer perspective what are all the different pieces they are going to be using is critical.
In other words, if you’re focused on a particular functional capability and that’s the only thing that you are aware of and that’s the only thing that you tackle, you’re doing your customer a disservice. There are too many moving pieces for any one vendor to tackle them all. So it’s going to be critically important that you’re partner-friendly, project-friendly and so on and integrate well and play nicely with others.
Governor: But also, don’t let a crisis go to waste. IT ops has budget, but they’re also always getting a kick in the teeth. Anything that goes wrong is their fault, even if it’s someone else’s. The simple fact is that we’re in an environment where organizations, as I’ve said, are thinking that the threat and opportunity posture has changed. It’s time to invest in this.
A good example of this would be that we always talk about standardization, but then nobody wants to give us the budget to do that. One of the things that we’ve tended to see in these web-native companies and how they manage operations and so on is that they’ve done an awful lot of standardization on what the infrastructure looks like. So there is an opportunity here. It’s a threat and an opportunity as well.
Gardner: I’ve been speaking with a few users, and there are a couple of rationales from them on what accelerates DevOps adoption. One of them is security and compliance, because the operations people can get more information back to the developers. Developers can insist that security gets baked in early and often.
The other one is user experience. The operations side of the house now has the data from the user, especially when we talk about mobile apps and smaller customer-facing applications and web apps. All that data is now being gathered. What happens with the application can be given back to development very quickly. So there is a feedback loop that’s compressed.
What do you think needs to happen in order for the incentives to quicken the adoption of DevOps from the perspective of security, user experience, and feedback loops of shared data?
Governor: That’s such a good question. It’s going to remain an ongoing challenge. The simple fact is that, as I said about the retail and the mobile app, different parts of the business have different goals. Finance doesn’t have the same goals as sales, and sales does not have the same goals as marketing in fact.
Within IT, there are different groups that have had very different key performance indicators (KPIs), and that’s part of the discussion. I think you are absolutely right to bring that up, understanding what are the metrics that we should be using, what are the new metrics for success? Is it the number of new products or changes to our application code that we can run out?
We’re all incredibly impressed by Etsy and Netflix because they can make all of these changes per day to their production environment. Not everybody wants to do that, but I think it’s what these KPIs are.
It might be, as Stephen had mentioned, if previously we were waiting six months to get access to server and storage, and we get that down to a minute or so, it’s pretty obvious that that’s a substantive step forward.
You’re absolutely right to say that it is about the data. When we began on this transition around agile and so on, there was a notion that those guys don’t care about data, they don’t care about compliance. The opposite is true, and there has been a real focus on data to enable the developer to do better work.
In terms of this shift that we’re seeing, there’s an interesting model that, funnily enough, HPE has begun talking about, which is “shifting left.” What they mean by that is taking that testing earlier into the process.
We had been in an environment where a developer would hand off to someone else, who would hand off to someone else, at every step of the way. The notion that testing happens early and happens often is super important in this regard.
Gardner: Continuous delivery and service virtualization are really taking off now. I just want to give Stephen an opportunity to address this alignment of interests, security, user experience, and shared data, and thoughts about how organizations should encourage adoption using these aligned interests.
O’Grady: I can’t speak to this query angle as much. In other words, there are aspects to that, particularly when we think about configuration management and the things that you can do via automation and so on.
The big one for me is user experience, and that to me is where a lot of the DevOps movement has come from. What we’ve found out is that if you want to deliver an ideal experience via an application to 100 people or 1,000 people, that’s not terribly difficult, and what you are using infrastructure-wise to address that is also not a sort of huge challenge.
On the flip side, you start talking millions and tens of millions of users, hundreds of millions of users potentially, then you have a completely different set of challenges involved. What we’ve seen from that is that the infrastructure necessary to deliver quality experiences, whether you’re Netflix, Facebook, or Google, or even just a large bank, that’s a brand-new challenge.
Then, when you get into not just delivering a quality experience through a browser, but delivering it through a mobile application, this encourages and, in fact, necessitates a series of architectural changes to scale out and all these other sort of wonderful things that we talk about.
Then, if we’re dealing with tens of thousands or hundreds of thousands of machines, instead of a handful of very, very large ones, we need a different approach, and that different approach in many respects is going to be DevOps. It’s going to be taking a very hands-on developer approach to traditional operational tasks.
Governor: But security is definitely an elephant stomping around the room. There’s no question. The feedback loop around DevOps has not been as fixated on security as it might be.
Quite frankly, developers are about getting things done and this is the constant challenge, ops, security, and so on. Look at Docker. Developers absolutely love it, but it didn’t start in a position of how do we make this the most secure model you could ever have for application deployment.
There are some weird people who started to use the word DevOps(Sec), but there are a lot of unicorns and rainbows and there is going to be a mess that needs clearing up. Security is something that we generally don’t do that well.
On the other hand, as I said, we’re less concerned with stability, and on the security side, it does seem like. Look at privacy. We all gave up on that, right?
Gardner: I suppose. Let’s not give up on security though.
Governor: Well, those things go together.
Gardner: They do.
Need to step up
Governor: Certainly, the organizations that would claim to be really good at security are the ones that have been leaving all of their customers’ details on a USB stick or on a laptop. The security industry has not done itself many favors. They need to step up as much as developers do.
Gardner: As we close out, maybe we can develop some suggestions for organizations that create a culture for DevOps or put in place the means for DevOps. Again, speaking with a number of users recently, automation and orchestration come to mind. Having that in place means being able to scale, to be able to provide the data back, monitoring, and data from a big-data perspective across systems to pan IT data, and the ability to measure that user experience. Any other thoughts about what you as an organization should put in place that will foster a DevOps mentality?
Governor: There are a couple of things. One thing you didn’t mention is pager duty. It’s a fact that somebody is going to get called out to fix the thing, and it’s about individuals taking responsibility. With that responsibility, give them a higher salary. That’s an interesting challenge for IT, because they’re always told, here are a bunch of tools that enable the Type As to get stuff done.
As to your point about whether this is a cultural shift or a product shift, the functional areas you mentioned are absolutely right, but as to the culture, just what’s important is to just get out and start spending time reading the stuff that the web companies are doing and sharing.
If you look at Etsy or Netflix, they’re not keeping this close to their chest. Netflix, in fact, has provided the tools it uses to improve stability through Chaos Monkey. So there’s much more sharing, there’s much more there, and the natural thing would be to go to your developer events. They’re the people building out this new culture. Embed yourself in this developer aesthetic, where GitHub talks about “optimizing for developer joy.” Etsy is about “Engineering Happiness.”
Gardner: Stephen, what should be in place in organizations to foster better DevOps adoption?
O’Grady: It’s an interesting question. The thing that comes to mind for me is a great story from Adrian Cockcroft, who used to be with Netflix. We’ve talked about him a couple of times. He’s now with Battery Ventures, and he gives a very interesting talk, where he goes out and talks to executives and senior technology executives from all of these Fortune 500 companies.
One of the things he get asked over and over and over is, “Where do you find engineers like the ones that work at Netflix? Where do we find these people that can do this sort of miraculous DevOps work?” And his response is, “We hired them from you.”
The singular lesson that I would tell all the organizations is that somewhere in your organization probably are the people who know how this stuff works and want to get it done. A lot of times, it’s basically just empowering them, getting out of the way and letting the stuff happen, as opposed to trying to put the brakes on all the time.
You may also be interested in:
- How INOVVO delivers big data network analysis for greater mobile user loyalty
- DevOps by design–A practical guide to effectively ushering DevOps into any organization
- Need for Fast Analytics in Healthcare Spurs Sogeti Converged Solutions Partnership Model
- HPE’s composable infrastructure sets stage for hybrid market brokering role
- Nottingham Trent University Elevates Big Data’s role to Improving Student Retention in Higher Education
- Forrester analyst Kurt Bittner on the inevitability of DevOps
- Agile on fire: IT enters the new era of ‘continuous’ everything
- Big data enables top user experiences and extreme personalization for Intuit TurboTax
- Feedback loops: The confluence of DevOps and big data
- IoT brings on development demands that DevOps manages best, say experts
- Big data generates new insights into what’s happening in the world’s tropical ecosystems
- DevOps and security, a match made in heaven
- How Sprint employs orchestration and automation to bring IT into DevOps readiness