OpenShift shops chart a course for advanced IT automation
IT teams tackled containerization, developer self-service platforms and infrastructure provisioning automation in 2018. So, what's next?
BOSTON -- Enterprise IT shops covered significant ground in IT automation in 2018, but they don't plan to slow the rapid pace of change anytime soon.
IT pros at Red Hat Summit here this week discussed their milestones in digital transformation in the last year, from initial production container workloads to day-to-day use of DevOps app delivery pipelines and a transition to microservices. Their to-do list remains long, however, with plans to bring legacy apps and stateful workloads into the IT automation fold, explore serverless architectures, and consider ways to apply advanced data analytics to IT infrastructure management.
The first item on the agenda for one major retailer will be to evaluate how it can apply Operators, a feature of Red Hat OpenShift 4.1, to ongoing IT operations. The company already automated infrastructure setup in 2017 and 2018, with a combination of Google Cloud Platform, Ansible, Terraform and OpenShift 3.9.
"A lot of the workloads we run in OpenShift are stateless apps, and we're looking forward to moving more of our stateful apps into the platform," said Ritch Houdek, senior vice president of technology at Kohl's, based in Menomonee Falls, Wis., in a keynote presentation. "We've also done a great job automating the day-one stuff, but there's great opportunity to automate more of the day-two things, to integrate more with our messaging and database systems."
Today, containerize all apps and improve day-two ops
Legacy app modernization, IT automation beyond installation tasks and stateful container apps were popular topics among Red Hat Summit attendees this week. French commercial aircraft manufacturer Airbus plans to integrate OpenShift with its ServiceNow IT management system for ongoing infrastructure updates and refresh legacy Java applications.
"We have an old J2EE platform that is very much end of life, and we need to migrate those applications to OpenShift," said Colin Richards, cloud and container architect for Airbus, in a presentation.
Red Hat introduced a Java framework in March 2019 called Quarkus, which IT analysts said could help enterprise customers and soon-to-be parent company IBM bring large portfolios of Java apps up to cloud-native speed. Quarkus is designed to make resource-intensive legacy Java apps run more efficiently in containers on Kubernetes.
Ritch HoudekSenior vice president of technology, Kohl's
"Young developers look at people talking about Java the way I used to look at people talking about COBOL," said Tom Petrocelli, analyst at Amalgam Insights in Arlington, Mass. "But there's a lot of it out there. And if those apps aren't slimmed down, customers won't see any value to running only one container per server because of Java's bulk."
German financial services company Deutsche Bank also plans to move legacy Windows applications to OpenShift in the fourth quarter of 2019, when Red Hat makes Windows container support generally available in version 4.3.
So far, the company's early container deployments have focused on Red Hat Enterprise Linux applications, which make up 43% of the company's overall workloads, said Adam Spencer, director at Deutsche Bank. To get to the internal PaaS team's goal of hosting 80% of the company's apps by 2022, a move to Windows apps on Kubernetes will be necessary.
Deutsche Bank will improve ongoing IT automation with the Kubernetes Federation v2 utility for federated cluster management as the bank's PaaS expands, but the project and OpenShift's support for it must mature first, Spencer said.
"When we get to OpenShift 4, we have to make federation available, because we have 27 clusters today, and we'll have 40 by the end of the year and continue to grow," he said. "We have to move our management and control plane up a level to manage those environments."
Deutsche Bank wants to manage multiple clusters as single entities, rather than apply changes to clusters individually. It also wants to control how applications migrate between clusters through infrastructure APIs.
Tomorrow, tackle serverless architectures and AIOps
OpenShift 4.1 introduced serverless features in tech preview that will be developed further in 2019's quarterly releases. Once IT pros get a better feel for a broader set of container apps, those serverless utilities are next on their docket to evaluate. They include the open source Knative project that links function as a service (FaaS) with Kubernetes, along with a Knative Azure Functions integration developed by Red Hat and Microsoft called Kubernetes-based event-driven autoscaling.
"It's looking like Knative is the bring-your-own-FaaS winner," said James Anderson, senior principal enterprise architect at Sabre Corp., a travel industry software provider based in Southlake, Texas. "We'll look at it down the road after we work on database as a service. But to add event-driven auto-scaling on premises, we have to get really good at our capacity management."
Some early adopter enterprises have already looked beyond application modernization and advanced IT automation, and they see a marriage between artificial intelligence techniques in data analytics and IT automation as a likely next step.
"We want to reduce the elbow grease [in IT incident response] so it's a little less manual and a little more automated," said Ken Finnerty, president of information technology at shipping company UPS, based in Atlanta. "It's an area we're very interested in and looking carefully at over the next year or two."
It's still very early in the evolution of AIOps and artificial intelligence in general, Finnerty said. Companies like his must solve two sets of problems: First, it must glean useful insights from IT data sets; second, it must apply those insights in a useful way to IT operations tasks. So far, UPS has brought its data science departments into its IT reporting structure to begin that process. It has also begun to consider various third-party AIOps tools, though Finnerty declined to name any specific tools under consideration.
Such AIOps tools vendors, meanwhile, won't include Red Hat, at least not in the near future. The company touted its integrations with partners such as ProphetStor and uses some data analytics in its Red Hat Insights service, but has no plans to offer its own AIOps products for automated incident response or AIOps features as part of OpenShift, a company spokesperson said.