Friday, August 17, 2018

New strategies emerge to stem the costly downside of complex cloud choices

The next BriefingsDirect hybrid IT management strategies interview explores how jerry-rigged approaches to cloud adoption at many organizations have spawned complexity amid spiraling -- and even unknown -- costs.

We’ll hear now from an IT industry analyst about what causes unwieldy cloud use, and how new tools, processes, and methods are bringing insights and actionable analysis to regain control over hybrid IT sprawl.

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript or download a copy.

Here to help us explore new breeds of hybrid and multicloud management solutions is Rhett Dillingham, Vice President and Senior Analyst at Moor Insights and Strategy. The discussion is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: What makes hybrid and multicloud adoption so complex?

Dillingham: Regardless of how an enterprise has invested in public and private cloud use for the last decade, a lot of them ended up in a similar situation. They have a footprint on at least one or multiple public clouds. This is in addition to their private infrastructure, in whatever degree that private infrastructure has been cloud-enabled and turned into a cloud API-available infrastructure to their developers.

Dillingham
They have this footprint then across the hybrid infrastructure and multiple public clouds. Therefore, they need to decide how they are going to orchestrate on those various infrastructures -- and how they are going to manage in terms of control costs, security, and compliance. They are operating cloud-by-cloud, versus operating as a consolidated group of infrastructures that use common tooling. This is the real wrestling point for a lot of them, regardless of how they got here.

Gardner: Where are we in this as an evolution? Are things going to get worse before they get better in terms of these levels of complexity and heterogeneity?

Dillingham: We’re now at the point where this is so commonly recognized that we are well into the late majority of adopters of public cloud. The vast majority of the market is in this situation. We’re going to get worse from an enterprise market perspective.

We are also at the inflection point of requiring orchestration tooling, particularly with the advent of containers. Container orchestration is getting more mature in a way that is ready for broad adoption and trust by enterprises, so they can make bets on that technology and the platforms based on them.

Control issues 

On the control side, we’re still in the process of sorting out the tooling. You have a number of vendors innovating in the space, and there have been a number of startup efforts. Now, we’re seeing more of the historical infrastructure providers invest in the software capabilities and turning those into services -- whether it’s Hewlett Packard Enterprise (HPE), VMware, or Cisco, they are all making serious investments into the control aspect of hybrid IT. That’s because their value is private cloud but extends to public cloud with the same need for control.

Gardner: You mentioned containers, and they provide a common denominator approach so that you can apply them across different clouds, with less arduous and specific work than deploying without containerization. The attractiveness of containers comes because the private cloud people aren’t going to help you deal with your public cloud deployment issues. And the public clouds aren’t necessarily going to help you deal with other public clouds or private clouds. Is that why containers are so popular?
Learn More About
HPE OneSphere
Dillingham: If you go back to the fundamental basis of adoption of cloud and the value proposition, it was first and foremost about agility -- more so than cost efficiency. Containers are a way of extending that value, and getting much deeper into speed of development, time to market, and for innovation and experimentation.

Containerization is an improvement geared around that agility value that furthers cloud adoption. It is not a stark difference from virtual machines (VMs), in the sense of how the vendors support and view it.

So, I think a different angle on that would be that the use of VMs in public cloud was step one, containers was a significant step two that comes with an improved path to the agility and speed value. The value the vendor ecosystem is bringing with the platforms -- and how that works in a portable way across hybrid infrastructures and multi-cloud -- is more easily delivered with containers.

There’s going to be an enterprise world where orchestration runs specific to cloud infrastructure, public versus private, but different on various public clouds. And then there is going to be more commonality with containers by virtue of the Kubernetes project and Cloud Native Computing Foundation (CNCF) portfolio.

That’s going to deliver for new applications -- and those lifted and shifted into containers -- much more seamless use across these hybrid infrastructures, at least from the control perspective.

Gardner: We seem to be at a point where the number of cloud options has outstripped the ability to manage them. In a sense, the cart is in front of the horse; the horse being hybrid cloud management. But we are beginning to see more such management come to the fore. What does this mean in terms of previous approaches to management?

In other words, a lot of organizations already have management for solving a variety of systems heterogeneity issues. How should the new forms of management for cloud have a relationship with these older management tools for legacy IT?

Dillingham: That is a big question for enterprises. How much can they extend their existing toolsets to public cloud?

A lot of the vendors from the private [infrastructure] sector invested in delivering new management capabilities, but that isn’t where many started. I think the rush to adoption of public cloud -- and the focus on agility over cost-efficiency -- has driven a predominance of the culture of, “We are going to provide visibility and report and guide, but we are not going to control because of the business value of that agility.”
The tools have grown up as a delivery on visibility but not the control of the typical enterprise private infrastructure approach, which is set up for a disruptive orientation to the software and not continuity.

And the tools have grown up as a delivery on that visibility, versus the control of the typical enterprise private infrastructure approach, which is set up for a disruptive orientation to the software and not continuity. That is an advantage to vendors in those different spheres. I see that continuing.

Gardner: You mentioned both agility and cost as motivators for going to hybrid cloud, but do we get to the point where the complexity and heterogeneity spawn a lack of insight and control? Do we get to the point where we are no longer increasing agility? And that means we are probably not getting our best costs either.

Are we at a point where the complexity is subverting our agility and our ability to have predictable total costs?

Growing up in the cloud 

Dillingham: We are still a long away from maturity in effective use of cloud infrastructure. We are still at a point where just understanding what is optimal is pretty difficult across the various purchase and consumption options of public cloud by provider and in comparing that to an accurate cost model for private infrastructure. So, the tooling needs to be in place to support this.

There has been a lot of discussion recently about HPE OneSphere from Hewlett Packard Enterprise, where they have invested in delivering some of this comparability and the analytics to enable better decision-making. I see a lot of innovation in that space -- and that’s just the tooling.

There is also the management of the services, where the cloud managed service provider market is continuing to develop beyond just a brokering orientation. There is more value now in optimizing an enterprise’s footprint across various cloud infrastructures on the basis of optimal agility. And also creating value from services that can differentiate among different infrastructures – be it Amazon Web Services (AWS) versus Azure, and Google, and so forth – and provide the cost comparisons.

Gardner: Given that it’s important to show automation and ongoing IT productivity, are these new management tools including new levels of analytics, maybe even predictive insights, into how workloads and data can best become fungible -- and moved across different clouds -- based on the right performance and/or cost metrics?

Is that part of the attractiveness to a multi- and cross-cloud management capability? Does hybrid cloud management become a slippery slope toward impressive analytics and/or performance-oriented automation?

Dillingham: We’ve had investment in the tooling from the cloud providers, the software providers, and the infrastructure providers. Yet the insights have come more from the professional services’ realm than they have from the tooling realm. That’s provided a feedback loop that can now be applied across hybrid- and multi-cloud in a way that hasn’t come from the public cloud provider tools themselves.
Learn More About
HPE OneSphere
So, where I see the most innovation is from the providers that are trying to address multi-cloud environments and best feed innovation from their customer engagements from professional services. I like the opportunity HPE has to benefit from their acquisitions of Cloud Technology Partners and RedPixie, and then feeding those insights back into [product development]. I’ve seen a lot of examples about the work they’re doing in HPE OneSphere in moving those insights into action for customers through analytics.

Gardner: I was also thinking about the Nimble acquisition, and with InfoSight, and the opportunity for that intellectual property to come to bear on this, too.

Dillingham: Yes, which is really harvesting the value of the control and insights of the private infrastructure and the software-defined orientation of private infrastructure in comparison to the public cloud options.

Gardner: Tell us about Rhett Dillingham. You haven’t been an IT industry analyst forever. Please tell us a bit about your background.

Dillingham: I’ve been a longtime product management leader. I started in hardware, at AMD, and moved into software. Before the cloud days, I was at Microsoft. Next I was building out the early capabilities at AWS, such as Elastic Compute Cloud (EC2) and Elastic Block Store (EBS). Then I went into a portfolio of services at Rackspace, building those out at the platform level and the overall Rackspace public cloud. As the value of OpenStack matured into private use, I worked with a number of enterprises on private OpenStack cloud deployments.

As an analyst, I support project management-oriented, consultative, and go-to-market positioning of our clients.

Gardner: Let’s dwell on the product management side for a bit. Given that the market is still immature, given what you know customers are seeking for a hybrid IT end-state, what should vendors such as HPE be doing in order to put together the right set of functions, processes, and simplicity -- and ultimately, analytics and automation -- to solve the mess among cloud adoption patterns and sprawl?

Clean up the cloud mess 

Dillingham: We talked about automation and orchestration, talked about control of cost, security, and compliance. I think that there is a tooling and services spectrum to be delivered on those. The third element that needs to be brought into the process is the control structure of each enterprise, of what their strategy is across the different infrastructures.

Where are they optimizing on cost based on what they can do in private infrastructure? Where are they setting up decision processes? What incremental services should be adopted? What incremental clouds should be adopted, such as what an Oracle and an IBM are positioning their cloud offerings to be for adoption beyond what’s already been adopted by a client in AWS, Google, and Azure?
The third element that needs to be brought into the process is the control structure of each enterprise, of what their strategy is across the different infrastructures.

I think there’s a synergy to be had across those needs. This spans from the software and services tooling, into the services and managed services, and in some cases when the enterprise is looking for an operational partner.

Gardner: One of the things that I struggle with, Rhett, is not just the process, the technology and the opportunity, but the people. Who in a typical enterprise IT organization should be tasked with such hybrid IT oversight and management? It involves more than just IT.

To me, it’s economics, it’s procurement, it’s contracts. It involves a bit more than red light, green light … on speed. Tell me about who or how organizations need to change to get the right people in charge of these new tools.

Who’s in charge?

Dillingham: More than the individuals, I think this is about the recognition of the need for partnerships between the business units, the development organizations, and the operational IT organization’s arm of the enterprise.

The focus on agility for business value had a lot of the cloud adoption led by the business units and the application development organizations. As the focus on maturity mixes in the control across security and compliance, those are traditional realms of the IT operational organization.

Now there’s the need for decision structure around sourcing -- where how they value incremental capabilities from more clouds and cloud providers is a decision of tradeoffs and complexity. As you were mentioning, of weighing between the incremental value of an additional provider and an incremental service, and portability across those.

What I am seeing in the most mature setups are partnerships across the orientations of those organizations. That includes the acknowledgment and reconciliation of those tradeoffs in long-term portability of applications across infrastructures – against the value of adoption of proprietary capabilities, such as deeper cognitive machine learning (ML) automation and Internet of Things (IoT) capabilities, which are some of the drivers of the more specific public cloud platform uses.

Gardner: So with adopting cloud, you need to think about the organizational implications and refactor how your business operates. This is not just bolting on a cloud capability. You have to rethink how you are doing business across the board in order to take full advantage.

Dillingham: There is wide recognition of that theme. It gets into the nuts and bolts as you adopt a platform and you determine exactly how the operations function and roles are going to be defined. It means determining who is going to handle what, such as how much you are going to empower developers to do things themselves. With the accountability that results, more tradeoffs are there for them in their roles. But it's almost over-rotation and focus to that out of recognition of it and lack of valuation of that more senior-level decision making in what their cloud strategy is.
Learn More About
HPE OneSphere
I hear a lot of cloud strategies that are as simple as, “Yes, we are allowing and empowering adoption of cloud by our development teams,” without the second-level recognition of the need to have a strategy for what the guidelines are for that adoption – not in the sense of just controlling costs, but in the sense of: How do you view the value of long-term portability? How do you value strategic sourcing and the ability to negotiate across these providers long-term with evidence and demonstrable portability of your application portfolio?

Gardner: In order to make those proper calls on where you want to go with cloud and to what degree, across which provider, organizations like HPE are coming up with new tools.

So we have heard about HPE OneSphere. We are now seeing HPE’s GreenLake Hybrid Cloud, which is a use of HPE OneSphere management as a service. Is that the way to go? Should we think of cloud management oversight and optimization as a set of services, rather than a product or a tool? It seems to me that a set of services, with an ecosystem behind them, is pretty powerful.

A three-layer cloud 

Dillingham: I think there are three layers to that. One is the tool, whether that is consumed as software or as a service.

Second is the professional consultative services around that, to the degree that you as an enterprise need help getting up to speed in how your organization needs to adjust to benefit from the tools and the capabilities the tools are wrangling.

And then third is a decision on whether you need an operational partner from a managed service provider perspective, and that's where HPE is stepping up and saying we will handle all three of these. We will deliver your tools in various consumption models on through to a software-as-a-service (SaaS) delivery model, for example, with HPE OneSphere. And we will operate the services for you beyond that SaaS control portal into your infrastructure management, across a hybrid footprint, with the HPE GreenLake Hybrid Cloud offering. It is very compelling.
HPE is stepping up with OneSphere and saying they will handle delivery of tools, SaaS models, and managed cloud services -- all through a control portal.

Gardner: With so many moving parts, it seems that we need certain things to converge, which is always tricky. So to use the analogy of properly intercepting a hockey puck, the skater is the vendor trying to provide these services, the hockey puck is the end-user organization that has complexity problems, and the ice is a wide-open market. We would like to have them all come together productively at some point in the future.

We have talked about the vendors; we understand the market pretty well. But what should the end-user organizations be starting to do and think in order for them to be prepared to take advantage of these tools? What should be happening inside your development, your DevOps, and that larger overview of process and organization in order to say, “Okay, we’re going to take advantage of that hockey player when they are ready, so that we can really come together and be proficient as a cloud-first organization?”

Commit to an action plan

Dillingham: You need to have a plan in place for each element we have talked about. There needs to be a plan in place for how you are maturing your toolset in cloud-native development… how you are supporting that on the development side from a continuousintegration (CI) and continuous delivery (CD) perspective; how you are reconciling that with the operational toolset and the culture of operating in a DevOps model with whatever degree of iterative development you want to enable.

Is the tooling in place from an orchestration and development capability and operations perspective, which can be containers or not? And that gets into container orchestration and the cloud management platforms. There is the control aspect. What tooling you are going to apply there, how you are going to consume that, and how much you want to provide it as a consultative offer? And then how much do you want those options managed for you by an operational partner? And then how you are going to set up your decision-making structure internally?

Every element of that is where you need to be maturing your capabilities. A lot of the starting baseline for the consultative value of a professional services partner is walking you through the decision-making that is common to every organization on each of those fronts, and then enabling a deep discussion of where you want to be in 3, 5, or 10 years, and deciding proactively.

More importantly than anything, what is the goal? There is a lot of oversimplification of what the goal is – such as adoption of cloud and picking of best-of-breed tools -- without a vision yet for where you want the organization to be and how much it benefits from the agility and speed value, and the cost efficiency opportunity.

Gardner: It’s clear that those organizations that can take that holistic view, that have the long-term picture in mind, and can actually execute on it, have a significant advantage in whatever market they are in. Is that fair?
Learn More About
HPE OneSphere
Dillingham: It is. And one thing that I think we tend to gloss over -- but does exist -- is a dynamic where some of the decision-makers are not necessarily incentivized to think and consider these options on a long-term basis.

The folks who are in role, often for one to three years before moving to a different role or a different enterprise, are going to consider these options differently than someone who has been in role for 5 or 10 years and intends to be there through this full cycle and outcome. I see those decisions made differently, and I think sometimes the executives watching this transpire are missing that dynamic and allowing some decisions to be made that are more short-term oriented than long-term.

Listen to the podcast. Find it on iTunes. Get the mobile app. Read a full transcript or download a copy. Sponsor: Hewlett Packard Enterprise.

You may also be interested in: