I was invited by the Tech Reckoning team, led by John and Kat Troyer, and facilitated by Amy White a couple weeks back to visit the Oracle campus in Redwood Shores, CA to learn about the Ravello Systems tool for Oracle Cloud.
Ravello has created an astounding piece of software that essentially allows for the creation of a duplicate model of a VMware or a KVM environment, with more hypervisor support to come. Essentially, the goal is to stand up virtual machines on either bare metal, or on a nested hypervisor based location, oddly, on Oracle’s cloud.
Seems to my initial thinking to be an incongruous relationship. VMware based VM’s sitting on Oracle? However, the interesting and unique thing is that these are native VMware guests, they’re able to run as native VMware VM’s. However, the incredible approach here is that you can essentially click, drag, and build an architectural model from the VMware environment as icon representations of objects in that infrastructure a multi-tiered workload build including switch elements, etc. that mirrors identically the environment already in place. By doing this, a Ravello user can model the portion of the virtual platform, and redeploy it easily. The idea is that a template or blueprint for that workload gets created and such, the deployment of a mirror of it regardless of the size of that environment becomes a trivial, replicable, and automated process. Using this tool, a segment of the environment, or even the entire virtual infrastructure being managed can be blueprinted and redeployed. They’ve even resolved the networking question. The virtual machine’s metadata is a portion of its image, so if that vm is part of the template, it takes on the same networking, IP address, routing, subnet, etc. rules initially in place. In order, then to accommodate for the physical networking components, like firewalls and load balancers, a virtual image of such would need to be implemented before the rules would function within a full virtual environment.
Conceptually, this feels very compelling. In a testing mode, for example. To be able to grab your entire environment, and without causing it any harm (of course it is a mirror image) test against it pieces like patching, or security testing, without actually harming the production environment, and being able to push that out as often or as freshly as necessary because that blueprint is a moment in time, is hugely significant. In other models, a test-dev mirror would need to be built, snapshotted, and then rebuilt for this kind of testing to take place.
This function alone could really solve so many problems to which an IT department may struggle. But, this is only the tip of the iceberg.
The vision for the product is that it be able to stand up a fully deployed production environment in the exact same way. So, for example, if an application has been created, one that requires many components, the ease of creating that blueprint (either from a click-and-drag approach, or also by leveraging the Restful API) and then simply deploying the blueprint thereby ensuring a perfect replica (or as many as necessary) of this model can be easily made and housed in the cloud is almost like the container model on steroids.
Pretty cool, huh?
But that’s not the only thing, because Ravello, as a product was already just about here a year ago, when they got acquired by Oracle. What then has happened between the acquisition and now? As Clay Magouyrk – VP Oracle Cloud Infrastructure told the group, the biggest task for the team since acquisition was to ensure that the Oracle Cloud as an underpinned infrastructure onto which to deploy Ravello based blueprints actually worked.
They opened a business office in Washington state, populated it with newly hired people who’d come from Google, Microsoft and Amazon, and built the platform from the ground up. Now over 900 people. All dedicated to the Oracle cloud and the integration of Ravello into it.
By all accounts, this year, while stealthy, was incredibly well spent. The platform worked quite well the first time I saw it, but now, with seamless integration into the Oracle Cloud is a truly impressive accomplishment. In addition, the ability to launch a fully native VMware implementation onto Oracle’s cloud platform impresses, and ensures that Oracle will have a truly robust platform for the full integration of enterprise architecture beyond the sole use-case of Oracle Databases, and data backup.
In what I see as a key expansion of the offerings Oracle has in their fresh cloud offerings, their ability to host native VMware workloads, while not necessarily ready for production workloads at this point, they’re certainly moving in that direction, and moving quickly.
Why is it not yet ready? Well, the VMware production layers require some level of fault tolerance, or at least DRS (Distributed Resource Scheduling), it’s clear that this is not yet in place. Should your workload not require something along these lines, then the ability to run these workloads works flawlessly. So, as a test bed for fully deployed applications, and however you’ve desired to run that testing, I see it as an amazing way to avoid the requirement of a sandbox.
Some of my peers have written awesome postings on the Ravello Blogger’s Day presentations.
Chris Wahl (@ChrisWahl) – http://wahlnetwork.com/2017/05/24/oracle-ravello-cloud-service/#more-14924
Max Mortillaro (@DarkAvenger) – https://www.kamshin.com/2017/05/oracle-ravello-cloud-service/
Amy Renee (@AmyEngineer) – https://amyengineer.com/2017/06/02/oracle-ravello-blogger-day-2017/
Ather Beg (@AtherBeg) – http://atherbeg.com/2017/05/31/oracle-ravello-blogger-day-part-2-ravello/
James Green (@JDGreen) – http://www.actualtech.io/can-oracle-build-viable-public-cloud/
Keith Townsend (@CTOAdvisor) did a great Facebook Live presentation, also on Periscope, which presented his views as well. https://t.co/8DgihxsrZX