In February 2016 Oracle announced it acquired Israel based Ravello Systems. Ravello delivers what is called a cloud hypervisor. Customers can run their VMware and KVM virtual machines without any modification to the networking and storage configuration on Amazon AWS, Google EC2 and Oracle Cloud platforms.
Ravello basically delivers a ‘nested ESXi/KVM as a service’. Customers wanting to run their VMware/KVM workloads on a public cloud pay for the consumption of resources and the cloud hypervisor to Ravello. Ravello makes sure the cloud platform is paid.
Since the acquisition was announced many people were worried the development of Ravello would stall. It wouldn’t be for the first time great technology was lost after being bought by Oracle.
This does not seem to be the case for Ravello. To start with, Ravello added support for Oracle Cloud after the acquisition by Oracle.
Secondly, the service was renamed to ‘Oracle Ravello Cloud Service‘
And third, the Oracle Ravello Cloud Service is called by Deepak Patil , vice president product development for Oracle, as a competitive differentiator.
See this video for the statement and future of Ravello. At 11:29 into the interview Patil talks about Ravello.
So what problem does Ravello solves for customers?
The Oracle Ravello Cloud Service allows to lift and shift applications running in virtual machine on VMware ESXi or KVM to the public cloud without any modification to the virtual disk format or the networking configuration.
In many cases when workloads needs to be migrated to public cloud using lift and shift there are several challenges:
- the virtual disk format is not supported by the cloud vendor. For example when moving VMware based workloads using VMDK as virtual disk format, customers cannot just upload the VMDK to Amazon as this diskformat is not support. A V2V has to be performed which has some challenges.
- the networking configuration has to adjusted as the cloud provider does not support stretching the private IP-space used on-premises.
There are also challenges in transporting the data from on-premises to cloud. Depending on the amount of data and available bandwidth this can take up to weeks. However Ravello is not going to solve this issue.
The capabilities of Ravello are targeted at moving test/dev worksloads , running demo, Proof of Concept and learning on public cloud. I have not seen any scenario using Ravello for moving production workloads to the cloud. The overlay of the Ravello cloud hypervisor probably hurts performance too much for this use case.