BOSTON -- Container portability between clouds hit a new milestone this week, but for mainstream companies, the container paradigm shift is still easier to discuss than to implement.
Red Hat and Amazon Web Services (AWS) aim to alter the landscape of container portability between cloud infrastructures with new collaboration around OpenShift due out in the fall. The two companies, partners since 2008, said this new work will not only allow OpenShift to manage containers in AWS, but also bring management of* AWS offerings -- such as Simple Storage Service, Elastic Load Balancing and the Route 53 domain-name-system service -- on premises for Red Hat users.
With this effort, kicked off here at the Red Hat Summit, Amazon joins rival Microsoft Azure with stronger hybrid cloud offerings and increased support for Kubernetes container orchestration. Amazon has also pledged to work on Kubernetes development upstream with Red Hat, according to Red Hat officials.
Kubernetes edges closer to a de facto standard in the container orchestration market with this development -- Amazon had been the lone holdout among major public cloud providers to jump on the Kubernetes development bandwagon. This also represents major momentum around hybrid cloud as the end state for enterprise workloads, only some of which can be run outside the corporate firewall.
"This fundamentally shifts the [cloud] market," said Robert Stroud, analyst with Forrester Research. "It extends the reach of Amazon into on-premises services and offers highly regulated customers an opportunity to look at public cloud as an additional option."
Enterprises struggle to scale steep OpenShift learning curve
The trail that Red Hat and Amazon look to blaze is in the right direction, according to IT pros, but some enterprise shops are already slow to put OpenShift into practice, without the added step to span hybrid clouds with it.
One OpenShift user, LogistiCare Solutions LLC, a nonemergency medical transportation company headquartered in Atlanta, has had trouble finding employees with the right skills and experience to work with the platform, especially as its popularity grows among enterprises.
"We're competing with another healthcare company in Florida that also just switched to OpenShift," said Michael Quintero, enterprise solution architect for the company. A systems integrator set up OpenShift in LogistiCare's environment, but "internal IT still doesn't know how it works."
The integrator partner offered to train Quintero's employees, but the 40 or 50 senior IT employees had too much existing responsibility to prioritize the new tool, Quintero said. Now, some of them advocate a switch to Amazon's EC2 Container Service (ECS) instead. Integration between AWS and OpenShift that includes ECS would appeal to Quintero, but Red Hat officials this week said ECS is not included in the integration between the two companies at this point.
OpenShift spans a broad swath of new technologies, such as containers, software-defined storage and software-defined networks, but also requires skills in administering VMs and server hardware in on-premises deployments. For some organizations, the devil's in the ocean of details the platform entails.
Take Miles & More, a European frequent-flyer and loyalty program management company spun off from Lufthansa in 2015. The company's IT staff was tasked not only to move its portion of Lufthansa's infrastructure into owned facilities, but also break down a legacy app into microservices and get the OpenShift platform running, according to a Red Hat Summit presentation by Matthias Krohnen, project manager at Miles & More.
Along the way, the OpenShift deployment faltered and nearly failed due to the hardware setup on which the company tried to run it: three servers, each with two virtual CPUs and 8 GB of RAM. Miles & More brought in a Red Hat solutions architect who redesigned the hardware infrastructure to encompass three times as many servers: three master nodes, two infrastructure nodes and four workers.
AWS integration highlights ongoing OpenShift maturation
Red Hat cited a quickly expanding list of enterprise customers that have pushed containers into production with OpenShift, such as Barclays, Disney, Massachusetts General Hospital, Deutsche Bank and UnitedHealth Group, as evidence that the learning curve isn't insurmountable.
OpenShift has also come a long way from its early days, said Dietmar Fauser, vice president of architecture, quality and governance for Amadeus, a global travel technology company based in Madrid. When Amadeus first chose OpenShift and Kubernetes in 2013, tools such as Prometheus to monitor the Kubernetes application stack didn't exist yet. It took Amadeus until 2016 to get OpenShift into production as the technology developed, Fauser said, but the platform has matured considerably.
Dietmar Fauservice president of architecture, quality and governance, Amadeus
For example, the integration with public cloud services from AWS could potentially ease some infrastructure management pain, Fauser pointed out. Any sufficiently powerful enterprise software platform will require a lot of infrastructure to host it.
"It's a serious deployment -- you have to be conscious when you make choices and be ready to invest in people that understand the platform," he said. "It might be simpler for some corporations to go down the AWS OpenShift path, to help them structure OpenShift clusters."
In the meantime, Red Hat's work on OpenShift and Kubernetes is far from done. Amadeus is eager to get its hands on federated cluster management and integrated support for database clusters, which are developing upstream in OpenShift and Kubernetes, Fauser said. Both will be added to the platform in the next year or so, Red Hat officials said.
Amadeus would also like to see integrated application and infrastructure monitoring under OpenShift, Fauser said, but Red Hat is wary to attempt a single-pane-of-glass offering. These all-in-one constructs have been unsuccessful in the past for other types of infrastructure, and that equation doesn't change with the introduction of containers, said Brian Gracely, director of product strategy for OpenShift at Red Hat.
"What we see more in practice is, devs want certain things, and ops want certain things, and the best thing is to figure out how they can share data to troubleshoot better," Gracely said.
* Information changed after publication
Pivotal adds Kubernetes integration to Cloud Foundry
Kubernetes improves Docker portability with multicloud orchestration
Kubernetes addresses container security and complexity