At the core of Oracle’s hybrid cloud computing strategy is an approach to containers that relies on a management plane residing in the network rather than in the hypervisor.
At its CloudWorld event this week, Thomas Kurian, president of Oracle product development, said the company’s hybrid cloud revolves around a microservices-based runtime that can run on multiple cloud computing platforms. That runtime then can be linked back to an Oracle Virtual Cloud Network (VCN) that houses the management plane the company uses to manage hypervisors and bare-metal servers capable of hosting Docker containers.
At present, Oracle VCN is available in only a handful of its public cloud regions. But the goal is to extend Oracle VCN not only to every region of the Oracle public cloud, but also to private clouds based on the same software the company uses to manage its public cloud. Those private clouds can be deployed either on the Oracle cloud or in a local data center.
At a time when traditional enterprise IT vendors are looking to regain some competitive footing against rivals such as Amazon Web Services (AWS) and Microsoft, having a platform-neutral approach to containers may prove key. Both AWS and Microsoft have tied their management planes to the hypervisors they developed for their platforms. That works fine when containers are deployed on top of hypervisors, but as more organizations decide they want to be able to run hundreds of containers per server many more of them will opt to eliminate hypervisors in favor of bare-metal servers. Today, the average server running a hypervisor can effectively support only 12 to 36 containers.
None of this means hypervisors are going away anytime soon. But IT organizations will find themselves managing a mix of types of container deployments. Oracle is making the case for shifting the management plane into the network as means for allowing IT organizations to employ containers, hypervisors and bare-metal servers as they see fit.
It remains to be seen whether Oracle can get customers that already have made significant commitments to AWS and Microsoft Azure to switch. The company is betting that most enterprise customers have yet to standardize on a cloud platform. That leaves room to convince them that tying the management plane to a specific hypervisor is, in effect, locking them into a specific cloud platform. The cost of making that decision may not be apparent today. But, over time, it’s expected IT organizations will want the ability to move workloads across multiple clouds with as little as friction as possible. Having a highly portable microservices runtime is the first step to reducing that friction.
In the meantime, Oracle is pouring engineering resources into is cloud platforms. The company now has more than 1,000 engineers working on its cloud platforms, double last year’s number. And Oracle expects to double that number again in the coming year. Of course, throwing bodies at a problem doesn’t necessarily solve it. But in the case of Oracle, those additional engineering resources do reflect a level of seriousness about the cloud that previously was absent.