Today's enterprises are caught between the 'rock' of competitive disruption and the 'hard place' of maintaining and modernising the legacy applications that keep the show on the road. Somehow, CIOs need to keep the lights on while making room for innovation to drive the business forward. But how can enterprise apps be built and deployed more efficiently? The answer, according to Red Hat, is microservices, open-source container platforms and hybrid cloud infrastructure.

Ashesh Badani, VP and GM of OpenShift.
 Image: Red Hat
We talked to Ashesh Badani, vice-president and general manager of OpenShift at Red Hat, to discover how businesses can achieve this difficult juggling act using next-generation IT architecture.

Badani began by noting that digital transformation and DevOps may, like SOA before it, be the IT-journalistic "gift that keeps on giving", because, no matter how many articles are written on the subject, people still seek clarification.

"The shorthand I use for digital transformation is: IT organisations have this 80/20 rule, spending 80 percent of their dollars on existing apps and 20 percent on innovation -- digital transformation is saying 'how can I flip those two around, or at least get closer to reversing them?'"

Presumably the hardest part for enterprises is modernising those existing apps and reducing the 80 percent figure -- something that disruptive competitors aren't encumbered with. Does that tally with your experience of customer conversations?

"Sometimes I feel I might have a slightly biased view of the market, because, by definition, if I've showed up they [the customers] are already some way down the road. In Silicon Valley -- where you may have a neighbour who works for Snap and another who works for Uber -- there is a different mindset, but if I go and meet with a customer in the mid-west, say, their journey is very different. For example, the CIO of a large credit-card company said to me: 'I have 700 applications, and a dozen or so can be done in cloud-native style -- but what am I supposed to do with the other 680?' As we go down this path -- and this is something we at Red Hat are passionate about -- how do we make sure that the existing investments that companies have made are taken along on this digital transformation/DevOps journey?"

So how exactly does Red Hat help with the modernisation of enterprises' legacy app portfolios?

"I'll give you a tangible example from KeyBank, a regional bank in the US. They had this big monolithic digital banking application with quarterly update cycles and a multi-step approval process before any change could be made to the app. They said: 'This is crazy, this is not DevOps, we can't update a digital app once a quarter -- we need to be able to do it, if not daily, then weekly'."

"So they went through a process, asking: 'How can we take this monolithic app and see what parts of it can be broken into microservices; how can we introduce this notion of DevOps; how can we introduce more automation -- how do you have more trust in the system so not everyone has to sign off for the app to get pushed live?'"

"They did that for a few months, and we [Red Hat] provided the platform [OpenShift] to make all these things easier -- to run containers in a hybrid cloud, support developer productivity, manage it operationally, and so on. But ultimately the behavioural and process change needs to come from within the organisation -- although Red Hat can can help via our professional services team and our Open Innovation Labs."


IT leader's guide to making DevOps work

More and more organizations are jumping on the DevOps bandwagon and benefiting
from increased productivity and a smoother workflow. This ebook looks at
how you can get the most from the DevOps approach. Free for Tech Pro Research subscribers.

Would it be fair to say that the question facing many CIOs is no longer "Which workloads should I move to the cloud?", but rather "Which, if any, workloads should I keep on-premises?"

"I think, at this point in time, for any new workload, the legitimate thing for enterprises to consider is 'should I be running this myself, or should I think about the public cloud?' What we're finding is that customers are looking at a platform like OpenShift -- using containers as having that abstraction level, above clouds. Because the notion is, you can have a private cloud, and if you use a sufficiently smart container platform like OpenShift -- which uses Kubernetes technology for orchestration, failover and so on -- you can get a lot of cloud benefits by running that in-house. But you can take that same platform and run it in the cloud. For example, a customer like [travel technology company] Amadeus is running OpenShift on VMware, on OpenStack, on AWS, as well as Google Cloud Platform. The point for them is, how do you remove the dependency they've had for years between applications that are built and the underlying infrastructure they run on?"

"When you talk to Amadeus, they'll tell you that what they track pretty closely is something called the 'look-to-book ratio' -- how many searches take place, versus how many bookings and transactions happen. Search volume is increasing, so now you've got to deal with increased volume and throughput -- but the increase in latency is also huge. Let's say Amadeus is serving an airline in Japan and another in Argentina. Situating your platform and infrastructure in Germany, for example, to serve a query from Japan will simply add latency -- the physics of the equation are against you. Take too long in responding -- and we're measuring this stuff in milliseconds -- and the customer will go elsewhere. Amadeus's case is 'I'm developing the same app, but the underlying infrastructure will be different -- I should have the flexibility to move that app and add features regardless of the infrastructure.'

How easy is it for Red Hat to support different cloud vendors?

"It's not hard to support more clouds, the question is, how many can we reasonably manage? We have three different flavours of OpenShift: one is OpenShift Container Platform, which the customer can choose to deploy on bare metal, in a virtualised environment, in a private cloud like OpenStack, or in a public cloud like AWS. The creation of the application, and the management of the operations -- that's all the customer's responsibility. Typically, enterprises go for the Container Platform."

"We have another offering called OpenShift Dedicated, which today is available on AWS and Google Cloud Platform, and is in early testing on Azure. This is the implementation that we [Red Hat] manage on behalf of the customer: in that environment we take clusters of OpenShift -- dedicated instances within these public clouds -- and we manage the platform. Obviously the app comes from the customer, but we manage the updating, the security and so on. So businesses lacking the operational skills can say 'I'll give it to you, Red Hat'."

"The third offering is OpenShift Online, which is a more traditional SaaS-based product, run in a multitenant public cloud: you just show up with your code, and we take care of everything else. So a dev/test thing might happen in OpenShift Online, larger organisations will run the Container Platform themselves, and then a smaller sub-group that wants to go public cloud but has some constraints around privacy, regulation or governance might choose Dedicated."

We often hear statements to the effect that 'eventually everything will run in the public cloud'. Is that an idea you subscribe to?

"I think, as things play out, the lines are going to be so blurred between private and public cloud. Essentially, think of them as pools of resources, some of which happen to be yours, while others happen to be from a third party. In some sense, if you have enough abstraction, enough automation and the platform is smart enough, you could just put a policy in place that says 'run this workload in the cheapest place possible for compute' or 'this workload must run in the EU'.

In terms of OpenShift's competitive set, is the product where you want it to be? Gartner's 2016 aPaaS Magic Quadrant placed Red Hat in the box marked 'Visionaries', with Salesforce and Microsoft leading the pack...

"The question is, what market are you talking about? There's a market for public cloud, and that's primarily IaaS -- where you could argue there are really only three relevant vendors [AWS, Azure and Google Cloud Platform]. Then there's the so-called market for PaaS, which is focused on infrastructure and 'ops', whereas the next 'container application platform' should be one that serves DevOps. So where is the 'dev' part? I feel like we're in an in-between zone: I've been to enough customer conversations where we spend the first fifteen minutes talking about 'what is PaaS?' -- I try to get away from that and say 'Look, here are the problems we solve, let's see where you are on the journey, and how we can help.'

Red Hat's model is to take upstream open-source developments, curate them and add value in terms of things like support and security, and then package them up for enterprise deployment. So what's happening upstream in the OpenShift world that's interesting to you -- serverless computing?

"A good question, and we've spent some time thinking about that. I feel we made two pretty significant investments around three years ago. The first was spotting the adoption of Docker-based Linux containers. We said 'this is going to be really big' and were one of the first companies to say 'we're going to rearchitect a platform to support this', putting a bunch of engineers to work on it upstream. The next one, around the same time, was Kubernetes -- when you've got hundreds or thousands of containers running at scale, how are you going to orchestrate and manage them? We're the leading contributor to that after Google."

"That, if you will, gives you the guts -- the foundation -- of a platform. Now, looking forward, there's almost an unlimited number of places we can go. Serverless would be one. You could also think, where is the opportunity around data and analytics -- what investments could we make around that? That's changing so quickly: a few years ago everyone was talking about Hadoop, now they're talking about Spark -- what will they talk about tomorrow? AI and ML, so Tensorflow, that could be an area."

"We think about this curation aspect all the time, but we don't want to spread ourselves too thin and become a jack-of-all-trades. We're investing a lot of time and energy in enabling self service, and you'll see announcements at Summit about something called 'service broker APIs' for creating a service catalogue. More and more ISVs will be able to plug in that and provide services -- examples are nginx, a very popular web server, F5 Networks, who are moving towards doing more things in a software-defined fashion, and NetApp, who are talking about making more storage available. I don't think we need another fundamental platform bet -- the next thing is, how do we build around the ones we've already made."

What are the sectors or types of enterprise where Red Hat finds the most traction?

"Where we're being successful is with companies who say 'We're going to build new apps, but we need to figure out what to do with our existing applications.' For example, a large bank I met with this morning in London, is going to move around 80 percent of its apps to run containerised by 2020 -- and we're talking about 6,000 applications here. The other 20 percent they're going to keep in place, manage and maintain -- but they want to make sure there's a way for those to interact, whether it's by sharing data or actual application functionality. Being able to support existing, established, legacy -- call them what you will -- applications as well as newer ones, we've found has been really important."

What are the key pain points that CIOs are experiencing as they go through this IT modernisation process? What do you expect to hear when you talk to them these days?

"I feel like there's a confluence of trends that's happened. One is -- and it's still ongoing -- about the cloud in general: 'Is it too expensive for us to run and manage ourselves? And if I do decide on the cloud, which workloads, when and how?' But increasingly we see customers getting comfortable with the tradeoffs, experimenting heavily and moving production workloads. Then there's the whole trend around microservices: 'I've built applications a certain way for a long time -- should I fundamentally rethink that?'

Are skills shortages a factor -- do you find that enterprise IT departments are generally up to speed with building cloud-native apps?

"I don't know if skills is the biggest issue, versus just cultural change -- how do you bring together dev teams and ops teams who have been working almost in parallel in different roles for so long? You need top-down encouragement and bottom-up groundswell to get change to happen. That's maybe the single largest impediment."

Read more