You are here

You are here

State of the hybrid enterprise: What's next for Dev and Ops?

David Linthicum Chief Cloud Strategy Officer, Deloitte Consulting
Hiking stepping stones on grass

Just 10 years ago, IT’s main concerns were to keep the systems in the data center running, and deal with the two-year backlog of applications that needed to be built.  Today, those concerns are pretty much the same.  But IT now has an opportunity to leverage public and private clouds, which will help it approach the efficiency of DevOps.  The question is: Should enterprises take advantage of these new approaches and technologies?

The answer is not simple, but at least today it’s answerable.  The hybrid enterprise is here.  You can now balance on-premises systems, such as traditional systems and private cloud, with the exploding use of public clouds.  This transformation changed IT forever, and presented some new opportunities for most enterprises.

Here's a look at the state of today's hybrid enterprise and what's next for Dev and Ops

Common patterns in the hybrid enterprise

There are a few common patterns emerging. They include:

The partial automation of development

The partial automation of development is part of moving to DevOps processes and a DevOps organization.  This means that the automation of development activities has occurred.  However, there are missing pieces.

For instance, while many organizations have automated some testing, such as unit and regression testing, they have not automated performance and penetration testing, which are still done manually.  The need for humans to intervene means that the process is less efficient, and the developers are not able to quickly respond to the needs of the business.

The under use of public clouds

While you would think that most enterprises are moving quickly to public clouds based upon the interest in the major providers, only about 5 to 8 percent of workloads have actually made it to public IaaS clouds, which are platform analogs for most data centers.

The reasons for this slow progress are that there is yet to be an automated way to move workloads, and many applications need to be refactored (redesigned and rebuilt) to take advantage of the native features of the public clouds.  Thus, enterprises find themselves unable to take advantage of the scalability and flexibility of the public cloud because they are not yet scalable and flexible when it comes to application and data migration to the cloud.  The irony of this situation is not lost on IT leadership.

The lagging integration of development (Dev) and operations (Ops)

While DevOps is really about automation and coordination between development and operations, this aspect of this emerging trend does not seem to be coming true as fast as it should.

The reasons are many, but the people issues lead the way.  Development and operations have been islands inside of IT, operating independently and sometimes not getting along.  Thus, you’re bound to get resistance if you try to push them together without a solid plan and set of incentives that will drive success.

Systemic to all of these patterns is that moving to new modes of IT—whether the cloud or DevOps—requires a great deal of change within organizations, including the changing of hearts and minds.  Most practitioners and IT leads have come around to the benefits of these emerging technologies; some staffers have yet to buy-in.  Moreover, and perhaps more importantly, hybrid enterprises may not have the budgets to affect change at the rate desired.

Moving forward with the hybrid enterprise

So what’s next for Dev and Ops in the hybrid enterprise?  The focus will be around solving the issues raised above, and leveraging the hybrid approach to better meet the needs of the business.  Indeed, we’re seeing a few major trends already.

The pragmatic hybrid cloud

The growth of the “pragmatic hybrid cloud” is the first trend.  In many cases, enterprises opt to not take advantage of private cloud platforms and they leave workloads on traditional systems that exist on-premises, and they make those systems work and play with systems that are public cloud-based.

The use of this hybrid cloud approach means that we’ll have public clouds as a platform option to reduce the hardware and software footprint within the enterprises.  However, the initial vision of “hybrid” included workloads that are drag-and-drop portable from public to private clouds.  This approach means that vision won’t be a reality.

The fact is, workloads placed on a public cloud are not likely to move from that public cloud.  This considers the cost of migrating to, as well as off, the public cloud.  The workloads typically go through some major renovations to make them run efficiently on cloud and non-cloud platforms, and no one is anxious to throw away that investment.

Improved costs and agility

There are many advantages that the hybrid enterprise will see.  The ability to optimize cost efficiency is one of the biggest.  Having the public cloud option means that there is no longer a need to tie capital expense to all application workloads.  Application development no longer needs to include hardware and software.  Instead, the ability to instantly provision the resources that are needed provides a huge agility advantage.

In order to realize this advantage, workloads need to go through the on-premises to cloud-migration process, and that is where the latency occurs.  Thus, the focus should not only be on leveraging private and public clouds, but the ability to automate the development and operations processes that will get the applications on the most efficient platforms.  That’s the problem that most hybrid enterprises need to solve. 

Finding automation

The hybrid enterprise needs to focus on the automation of both development and operations.  This automation involves three factors.

1. On-prem workloads

First, the ability to automate the redevelopment or migration of existing on-premises workloads for migration to the public cloud, and sometimes private clouds.  This means setting up a DevOps organization, along with processes, retraining, etc., that will allow the enterprises to take advantage of a migration factory to make short work of most application migrations that need to occur.

Enterprises have gotten to an application-a-day, in some instances, moving from on premises systems to the cloud.  They apply emerging DevOps concepts, as well as leverage tools, to move applications through continuous development, continuous integration, continuous testing, and continuous deployment.

2. Agile app dev

Second, the ability to automate net-new application development, application changes, and other activities that allow applications to change or appear as needed by the business.  This is the essence of DevOps.

Most organizations evolve toward the ability to set up a migration factory that can move thousands of application workloads to the cloud, and automatically deal with any refactoring that needs to take place.  They use that as a jumping off point to more formal DevOps processes and automation that becomes the platform on which they can redevelop and redeploy at the speed their business needs. Moreover, they can deploy to either cloud or on-premises platforms.

However, the actual state of things is that most hybrid enterprises have yet to set up the basic migration factory yet, which is really DevOps with training wheels.  They understand the value, and typically they have the vision in place; they know they need to evolve, but this path is a difficult and expensive one.

3. The need for cost awareness

Also well-known are the business benefits, and the fact that costs go up and benefits go down the longer it takes to get the factory in place, as well as to achieve DevOps-lite and full DevOps.  The ability to understand this concept is easy—you just have to add up the cost of inefficiencies as they affect the business.  For example, a company that is unable to automate a new factory at the speed they need will cause production to be delayed for a much as a year.  That can cost up to $0.5 mil a day, in terms of lost revenue.

The largest cost is the difference between the as-is state of development backlogs and a fully agile business, one that can respond to all market and business demands in real time.  A fully agile business can access the compute resources that we need, on-demand, and build or change the applications that exist on those platforms.  By the way, a fully agile business does not yet exist.  However, a few innovators have come close, such as Uber and Netflix, that are both heavy DevOps and cloud users.

What does progress look like?

The hybrid enterprise is one of those problems that will need a stepwise progression, and most are still at the planning stage.  What should be the motivator here is the fact that, for most businesses, this is game-changing.  It offers the ability to provide a strategic advantage based on better usage of technology.

My advice is simple.  The new hybrid enterprises need to exploit the technology that they have already adopted.  The use of public clouds and private clouds, all working and playing well with traditional systems, is currently the norm.  Workloads that should move to the clouds is now priority one.

But that’s not all that needs to be done. The ability to automate development and operations needs to occur at the same time, and sometimes this needs to pre-date the move to cloud to facilitate the migration itself.  So, which came first, DevOps or cloud? 

The answer is DevOps.

Image credit: Flickr

Keep learning

Read more articles about: Enterprise ITCloud