Podcast: Manage your total cloud costs better. Emphasize the 'Ops' in DevOps
[Editor's note: This podcast was originally published on March 13, 2019.]
When it comes to operating in a hybrid cloud enivironment, the “Dev” and “Ops” sides of DevOps are often at odds: While the former seeks free rein to innovate and get things done, the latter acts as the guardrail, instituting policies and controls to ensure proper governance and security.
But, to gain the most economic advantage from hybrid cloud, the two sides need to find common ground—with operations taking the lead on cost optimization, says Daniel Newman, principal analyst at Futurum Research. That means more orchestration and coordination between the groups, supported by new composable cloud tools that give operations people more control while giving developers more visibility.
Listen to this Hewlett Packard Enterprise Voice of the Analyst podcast with Newman and Interarbor Solutions’ Dana Gardner to learn how Dev and Ops can build shared goals and help your organization achieve long-term payback from its cloud deployments.
Dana Gardner: Hello, and welcome to the next edition of the BriefingsDirect Voice of the Analyst podcast series. I’m Dana Gardner, principal analyst at Interarbor Solutions, your host and moderator for this ongoing discussion on the latest insights into successful digital transformation.
This IT management strategies interview explores new ways that businesses can gain the most control and economic payback from various cloud computing models. We’ll now hear from an IT industry analyst on how developers and IT operators can find newfound common ground to make hybrid cloud the best long-term economic value for their organizations.
Here to help explore ways a managed and orchestrated cloud lifecycle culture should be sought across enterprise IT organizations is Daniel Newman, principal analyst and founding partner at Futurum Research. Welcome, Daniel.
Daniel Newman: Hi, Dana. It’s good to be here.
Gardner: Daniel, many tools have been delivered over the years for improving software development in the cloud. Recently, containerization and management of containers has been a big part of that.
Now, we’re also seeing IT operators tasked with making the most of cloud, hybrid cloud, and multicloud around DevOps―and they need better tools, too.
Has there been a divide or lag between what developers have been able to do in the public cloud environment and what operators must be able to do? If so, is that gap growing or shrinking now that new types of tools for automation, orchestration, and composability of infrastructure and cloud services are arriving?
Out of the shadow, into the cloud
Newman: Your question lends itself to the concept of shadow IT. The users of this shadow IT find a way to get what they need to get things done. They have had a period of uncanny freedom.
But this has led to a couple of things. First of all, generally nobody knows what anybody else is doing within the organization. The developers have been able to creatively find tools.
Daniel Newman, Futurum Research
On the other hand, IT has been cast inside of a box. And they say, “Here is the toolset you get. Here are your limitations. Here is how we want you to go about things. These are the policies.”
And in the data center world, that’s how everything gets built. This is the confined set of restrictions that makes a data center a data center.
But in a developer’s world, it’s always been about minimum viable product. It’s been about how to develop using tools that do what they need them to do and getting the code out as quickly as possible. And when it’s all in the cloud, the end user of the application doesn’t know which cloud it’s running on—they just know they’re getting access to the app.
Basically, we now have two worlds colliding. You have a world of strict, confined policies—and that’s the “Ops” side of DevOps. You also have the developers who have been given free rein to do what they need to do, to get what they need to get done, done.
Get Dev and Ops to collaborate
Gardner: So, we need to keep that creativity and innovation going for the developers so they can satisfy their requirements. At the same time, we need to put in guardrails to make it all sustainable.
Otherwise, we see not a minimal viable cloud but out-of-control expenses, out-of-control governance and security, and difficulty taking advantage of both private cloud and public cloud, or a hybrid affair, when you want to make that choice.
How do we begin to make this a case of worlds collaborating instead of worlds colliding?
Newman: It’s a great question. We have tended to point DevOps toward “Dev.” It’s really been about the development, and the “Ops” side is secondary. It’s like capital “D,” lowercase “o.”
The thing is, we’re now having a massive shift that requires more orchestration and coordination between these groups.
You mentioned out-of-control expenses. I spoke earlier about DevOps and developers having the free rein to do what they need to do, put it where they need to put it—containers, clouds, tools, whatever they need—and just get it out because that’s what impacts their customers.
If you have an application where people buy things on the web and you need to get that app out, it may be a little more expensive to deploy it without the support of Ops, but you feel the pressure to get it done quickly.
Now, Ops can come in and say, “Well, you know, what about a flex consumption-based model, what about multicloud, what about using containers to create more portability?
“What if we can keep it within the constraints of a budget and work together with you? And, by the way, we can help you understand which applications are running on which cloud and provide you the optimal [aggregate cloud use] plan.”
Let’s be very honest, a developer doesn’t care about all of that.… They are typically not paid or compensated in any way that leads to optimizing on cost. That’s what the Ops people do.
Such orchestration, just like almost all larger digital transformation efforts, starts when you have shared goals. The problem is, they call it a DevOps group, but Dev has one set of goals and Ops has different ones.
What you’re seeing is the need for new composable tools for cloud services, which we saw at such events as the recent Hewlett Packard Enterprise Discover conference. They are launching these tools, giving the Ops people more control over things and, by the way, giving developers more visibility than has existed in the past.
There is a big opportunity [for better cloud use economics] through better orchestration and collaboration, but it comes down to the age-old challenges inside of any IT organization—and that is having the Dev and the Ops people share the same goals. These new tools may give them more of a reason to start working in that way.
Gardner: The more composability the operations people have, the easier it is for them to define a path that the developers can stay inside of without encumbering the developers.
We may be at the point in the maturity of the industry where both sides can get what they want. It’s simply a matter of putting that together—the chocolate and peanut butter, if you will. It becomes more of a complete DevOps.
But there is another part of this people often don’t talk about, and that’s the data placement component. When we examine the lifecycle of a modern application, we’re not just developing it and staging it where it stays static. It has to be built upon and improved; we are doing iterations, we are doing agile methods.
We also have to think about the data the application is consuming and creating in the same way. That dynamic data use pattern needs to fit into a larger data management philosophy and architecture that includes multicloud support.
I think it’s becoming DevDataOps, not just DevOps these days. The operations people need to be able to put in requirements about how that data is managed within the confines of that application’s deployment, yet kept secure and in compliance with regulations and localization requirements.
Newman: We’ve launched the DevDataOps category right now! That’s actually a really great point, because if you think about where does all that live—meaning IT orchestration of the infrastructure choices and whether that’s in the cloud or on premises―there has to be enough of the right kind of storage.
Developers are usually worried about data from the sense of what can they do with that data to improve and enhance the applications. When you add in elements like machine learning and artificial intelligence, that’s going to just up the compute and storage requirements. You have the edge and Internet of Things to consider now too for data. Most applications are collecting more data in real time. With all of these complexities, you have to ask, “Who really owns this data?”
Well, the IT part of DevOps, the Ops, typically worries about capacity and resources performance for data. But are they really worried about the data in these new models? It brings in that needed third category because the Dev person doesn’t necessarily deal with the data lifecycle. The need to best use that data is a business unit imperative, a marketing-level issue, a sales-level data requirement. It can include all the data that’s created inside of a cloud instance of SAP or Salesforce.
Just think about how many people need to be involved in orchestration to maximize that. Culturally speaking, it goes back to shared tools, shared visibility, and shared goals. It’s also now about more orchestration required across more external groups. So your DevOps group just got bigger, because the data deluge is going to be the most valuable resource any company has. It will be, if it isn’t already today, the most influential variable in what your company becomes.
You can’t just leave that to developers and operators of IT. It becomes core to business unit leadership, and they need to have an impact. The business leadership should be asking, “We have all this data. What are we doing with it? How are we managing it? Where does it live? How do we pour it between different clouds? What stays on premises and what goes off? How do we govern it? How can we have governance over privacy and compliance?”
I would say most companies really struggle to keep up with compliance because there are so many rules about what kind of data you have, where it can live, how it should be managed, and how long it should be stored.
I think you bring up a great point, Dana. I could probably rattle on about this for a long, long time. You’ve just added a whole new element to DevOps, right here on this podcast. I don’t know that it has to do with specifically Dev or Ops, but I think it’s Dev+Ops+Data—a new leadership element for meaningful digital transformation.
Gardner: We talked about trying to bridge the gap between development and Ops, but I think there are other gaps, too. One is between data lifecycle management―for backup and recovery and making it the lowest cost storage environment, for example. Then there is the other group of data scientists who are warehousing that data, caching it, and grabbing more data from outside, third-party sources to do more analytics for the entire company. But these data strategies are too often still divorced.
These data science people and what the developers and operators are doing aren’t necessarily in sync. So, we might have another category, which would be Dev+Data+DataScience+Ops.
Add data analytics to the composition
Newman: Now we’re going four groups. You are firstly talking about the data from the running applications. That’s managed through pure orchestration in DevOps, and that works fine through composability tools. Those tools provide IT the capability to add guardrails to the developers, so they are not doing things in the shadows but instead doing things in coordination.
The other data category is that bigger analytical data. It includes open data, third-party data, and historical data that’s been collected and stored inside of instances of enterprise resource planning apps and customer relationship management apps for 20 or 30 years. It’s a gold mine of information. Now we have to figure out an extract process and incorporate that data into almost every enterprise-level application that developers are building. Right now, Dev and Ops don’t really have a clue what is out there and available across that category because that’s being managed somewhere else, through an analytics group of the company.
Gardner: Or, developers will have to create an entirely different class of applications for analytics alone, as well as integrating the analytics services into all of the existing apps.
Newman: One of the HPE partners I’ve worked with in the past, SAS, and companies such as SAS and SAP, are going to become more closely aligned with infrastructure. Your DevOps is going to become your analytics Ops, too.
Hardware companies have built software apps to run their hardware, but they haven’t been historically building software apps to run the data that sits on the hardware. That’s been managed by the businesses running business intelligence software, such as the ones I mentioned.
There is an opportunity for a new level of coordination to take place at the vendor level, because when you see these alliances and you see these partnerships, this isn’t new. But seeing it done in a way that’s about getting the maximum amount of usable data from one system into every application, that’s futuristic, and it needs to be worked on today.
Gardner: The bottom line is that there are many moving parts of IT that remain disjointed. But we are at the point now with composability and automation of getting an uber view over services and processes to start making these new connections―technically, culturally, and organizationally.
What I have seen from HPE around the HPE Composable Cloud vision moves a big step in that direction. It might be geared toward operators, but ultimately it’s geared toward the entire enterprise and gives the business an ability to coordinate, manage, and gain insights into all these different facets of a digital business.
Companies right now still struggle with the resources to run multicloud. They tend to have maybe one public cloud and their on-premises operations. They don't know which is the best cloud approach because they are not getting the total information.
Newman: We’ve been talking about where things can go, and it’s exciting. But let’s take a step back.
Multicloud is a really great concept. Hyperconverged infrastructure, it’s all really nice, and there has been massive movement in this area in the last couple of years. Companies right now still struggle with the resources to run multicloud. They tend to have maybe one public cloud and their on-premises operations. They have their own expertise, and they have endless contracts and partnerships.
They don’t know which is the best cloud approach because they are not necessarily getting that total information. It depends on all of the relationships, the disparate resources they have across Dev and Ops, and the data can change on a week-to-week basis. One cloud may have been perfect a month ago, yet all of a sudden, you change the way an application is running and consuming data, and it’s now in a different cloud.
What HPE is doing with HPE Composable Cloud takes the cloud plus composable infrastructure and, working through HPE OneSphere and HPE OneView, brings them all into a single view. We’re in a software and user experience world.
The tools that deliver the most usable and valuable dashboard-type of cloud use data in one spot are going to win the battle. You need that view in front of you for quick deployment, with quick builds, portability, and container management. HPE is setting itself in a good position for how we do this in one place.
Give me one view, give me my one screen to look at, and I think your Dev and Ops—and everybody in between―and all your new data and data science friends will all appreciate that view. HPE is on a good track, and I look forward to seeing what they do in the future.
Gardner: I’m afraid we’ll have to leave it there. We have been exploring new ways that businesses can gain the most control and economic payback from various cloud computing models. And we’ve learned how developers and IT operators can find newfound common ground to make hybrid cloud the best long-term economic value for their organizations.
Please join me in thanking out guest, Daniel Newman, principal analyst and founding partner at Futurum Research. Thank you, sir.
Newman: Great to be with you, Dana.
Gardner: And a big thank you as well to our audience for joining this BriefingsDirect Voice of the Analyst IT management strategies interview. I’m Dana Gardner, principal analyst at Interarbor Solutions, your host for this ongoing series of Hewlett Packard Enterprise-sponsored discussions.
Thanks again for listening. Please pass this along to your IT community and do come back next time.
Managing total cloud costs: Lessons for leaders
- The “Dev” and “Ops” sides of DevOps have different goals in hybrid cloud environments: While developers look to cloud to get the job done, operations is charged with keeping the costs under control.
- New composable cloud tools can help bring the two sides together, enabling greater orchestration and coordination.
- Dynamic data use also needs to fit into the multicloud management equation. That may mean a new organizational category―DevDataOps―whereby operations has a hand in how data within applications is managed, secured, and kept in compliance with regulations.
This article/content was written by the individual writer identified and does not necessarily reflect the view of Hewlett Packard Enterprise Company.