Oracle/Ravello Blogger Day 2 – Update
This past week I had the wonderful opportunity to attend Oracle’s Ravello VMware Blogger Day. This will be my second time attending this event and I’m most excited to hear about how Oracle’s cloud and cloud management vision last year has been executed on. Oracle and Ravello, you’ve had a whole year to execute, time to show off. In all 29 bloggers, a couple of Oracle customers, and several of the talented folks at Oracle joined together to discuss the progress that’s been made since we last met about 10 months ago.
Oracle Cloud Infrastructure (OCI) – Focused on the Enterprise
To kick off the day we reviewed the growth of OCI. Oracle admits that they’re behind other hyper-scale cloud providers with service richness but with hundreds of new features released in the last year, they’re making progress on gaining parity with larger cloud players. Will this level of velocity be enough to catch the Amazon’s and the Microsoft’s of the world? Time will tell, but there is another play in Oracle’s playbook that seems more interesting to me.
Part of the overall Oracle cloud execution vision relies on executing in the Enterprise space. Oracle’s vision is that customers should be able to run their entire business on the Oracle Cloud. To accomplish this feat, Oracle is placing a lot of effort in providing Enterprises the tools and mechanisms that they need to manage a cloud platform. From the infrastructure up, the Oracle cloud has been designed with the intent of being able to provide SLA bounded services to the enterprise.
Oracle’s infrastructure provides resiliency and availability from the start. Regions are composed of availability domains (think data centers connected by dark-fiber) not unlike Amazon’s configuration. Oracle planned a non-blocking network (think zero oversubscription) to guarantee performance that Enterprise customers demanded. All networking in the Oracle cloud is virtualized off of the box in the Oracle data centers meaning that updating networking within the data center (management plane tasks) can scale appropriately. Another advantage of this method is the ability to natively support bare-metal or virtualized workloads with the same networking bump in the wire shim. Personally I love to see how each cloud vendor has approached this problem… Amazon surely has a method to deal with these same sorts of problems – See: Another Day, Another Billion Packets.
Do you want a customer’s crown jewels? Does your service have an SLA?
SLA’s matter to enterprises. When SAP goes down at a manufacturer, or an EMR system at a Hospital goes down bad things happen, companies lose the ability to make money or patients lose systems critical to ensure that they are receiving the best healthcare. It’s very clear that Oracle understands this and also understands that these businesses want guarantees around their services hosted in the Oracle Cloud. A bit of a key differentiator for Oracle is that they’re aligning to provide SLA’s for their Storage, Network, and Compute performance and also manageability, not simply uptime. This takes what is common in the Microsoft and AWS and extends them to support performance as well as the ability for administrators to access and support the management of these assets. A bit of insurance goes a long way towards helping new cloud customers gain confidence in the platform. This is surely a play for the enterprise.
A strong understanding of how Enterprises want to pay for cloud consumption
While pay-as-you-go options for Oracle cloud exist, most enterprises are still very squeamish about the potential for out of control public cloud bills. In my day job this is one of the guardrails that we always work with customers to put in place. Oracle has a pretty interesting model here around invoicing allowing them to sell a bucket of access to Oracle Cloud. Where possible Oracle is also trying to eliminate charges that seem to add up out of nowhere… For example, networking egress isn’t charged when a customer uses Oracle Fast Connect. In the billing space, smart play Oracle.
Challenges that I see
Oracle is behind. Very far behind of the other hyper-scale cloud providers. I’m glad to see the progress that’s been made, but with other vendors focusing on meaningful service offerings that are close to developers hearts and minds (Artificial Intelligence / Machine Learning / Data Analytics) I still worry about Oracle’s ability to catch up in the table stakes arena while working on developing more meaningful services. I believe the success of the Oracle cloud will hinge on the ability of the Oracle Sales team to leverage existing relationships and land existing Oracle customers. This, in fact, may be the key. I shouldn’t underestimate the number of businesses that run their world on Oracle software. Telling these customers that we can host those services natively in the Oracle cloud is a nice choice for many who don’t want to worry about how to re-platform these complex systems. Clearly even with an easy button to onboard apps to the cloud, Oracle will still need to build out the ecosystem support and integration partners spaces, an area that’s come a long way in the last few months but still has a long way to go. Regarding services that are beyond the public cloud basic food groups ( things generally defined as supporting IaaS), I only bring this up because more and more of my customers are abandoning running Virtual Machines/instances at all in favor of tying services together to build stacks in the public cloud providers. Even the Oracle team called this out as something that they’re thinking about. I’d double down on thinking about it.
In the last 10 months, the Ravello development team has been busy. Initially, Ravello used binary translation to provide a decent level of performance on top of the HVX hypervisor running on any VM. The Oracle Ravello team has further invested in Virtualization Extensions that improve performance by in some cases 10x making running an actual production workload a possibility (Kudos here to the Ravello team). Along with this, Ravello now supports running HVX natively on top of Oracle Bare Metal Instances (we were told this was coming last time, it’s here now). Is this going to beat out VMware anytime soon? Probably not, but hey there are parts of this market that are going to erode over time and Oracle seems primed to try to take a bite out of that pie. I can also see Ravello play an interesting role in helping companies re-platform easily to the Oracle cloud and perhaps that’s the bigger play here.
Interesting Use Case for development with Ravello
Development use cases were presented as a part of the overall session. One that did make sense to me that was presented was the ability to run Amazon ami’s along with VMware vmdk’s natively for the development and testing of software that’s delivered as virtual appliances. One platform to provision and test all of those options seems to be pretty interesting to me. I do like the ability to revert to snapshots for development/testing. I know this isn’t a new trick for working with software development but it’s one to note as the platform gives you that capability out of the box.
More to come
As I dive in head first into the OCI platform (thanks for the free access Oracle) I’ll plan on covering automation of the platform as well as some of the features that I see that are different from other cloud providers. In fact, I know that this event has been driven by the Oracle Ravello group, but for me, the most interesting bits were covering the Oracle Cloud details. Until then, thank you Oracle for having me out to be a part of the Oracle Blogger Day event and I’ll look forward to a bigger Oracle Cloud day in the future.