Inside Oracle’s cloud strategy

Oracle may be late to the public cloud market, but the database pioneer believes it has what it takes to make a dent in the Asia-Pacific region.

That is according to Chris Chelliah, group vice-president and chief architect for core technology and cloud at Oracle Asia-Pacific, who noted the “latent demand” in the region for a cloud service that supports demanding database and enterprise class workloads.

In an interview with Computer Weekly, Chelliah offers insights on the state of cloud adoption in the region and Oracle’s cloud strategy, including what it did to address the drawbacks of competing cloud services and build on its heritage in database management.

What are your observations about the state of cloud adoption in the Asia-Pacific region?

Chelliah: There are probably three things that we’re hearing from our customers in Asia-Pacific and around the world. One, most customers I speak to today have some form of cloud development environment or cloud strategy. So, there’s no need to preach to them about why they need to go the cloud.

However, most of them have told me that when they started their cloud journey, whether it was one year or five years ago, the intention was that over time, they would be able to move more of their on-premise workloads to the cloud to take advantage of the economics and agility of cloud – and at a faster pace.

The second thing we hear from customers is they don’t want to move everything to one particular cloud without an exit strategy. So, they’re looking at strategies to keep their options open across multiple suppliers. Lastly, customers are also seeing regulations, sometimes even network latency, shaping their cloud strategy.

How is Oracle responding to what you’ve just described, given the on-premise infrastructure is here to stay for varying reasons like performance requirements and data sovereignty?

Chelliah: You’re right that most workloads are still on-premise today for those reasons you mentioned. In view of that, we offer three things in our cloud strategy for customers.

First, one of the inhibitors about going to the cloud has been around security, so being five years later than our competitors in the market is a blessing in disguise because what we’ve done is that we’ve rebuilt our cloud ground up with security in mind – and that’s just not a marketing statement.

Every other cloud supplier uses the hypervisor between how they manage and isolate your workloads. Just in the past 12 months, there have been several hypervisor breaches and at some point, that dam is going to break open. The hypervisor breach is going to be a big problem in the next five years.

There are some fundamental differences in how we built our enterprise grade cloud. We split our cloud implementation into two, where we have the customer data in one network segment, and all the code managing that data in another physically isolated segment.

Fundamentally, we give our customers a bare metal computer which has no Oracle software on it. You can’t retrofit this because it’s literally a new design of the network fabric, and because we remove the hypervisor tax, we get a tremendous amount of performance on our cloud.

The second thing we built on top of that is tied very much back to our heritage in data management. We launched the Autonomous Database two years ago and we have since enhanced that to be everything, from data warehousing to transaction processing.

One of the inhibitors about going to the cloud has been around security, so being five years later than our competitors in the market is a blessing in disguise

But what very few people know is we’re offering that as a converged data management platform. It’s not just for relational data – you can have spatial data, graph data, JSON data and NoSQL data. All of that is available in the Autonomous Database in one single offering, running on our physically separated second generation Oracle Cloud Infrastructure (OCI).

What that means is that for those building new applications, you have the notion of data as a microservice. Any one of your pieces of data, whether it is relational, spatial or graph, is available for you to provision in an autonomous database, which is patched, backed up, secured, monitored, managed for you out of the box.

The third thing we offer as a differentiator is complete deployment choice. You can pick up any on-premise workload like a database or infrastructure service and run that on the public cloud, keeping in place all your security, operating, monitoring and management frameworks. We also have the Oracle Cloud at Customer offering where we bring our public cloud into the customer datacentres. Lastly, our relationships with Microsoft and VMware give our joint customers full ability to deploy their applications on an enterprise grade, secure cloud.

That’s really our story. We’ve done what we said we’re going to do about two years ago. We’ve switched on 17 regions in the past year. We now operate 21 regions globally and will have 36 regions by the end of 2020. That is by far a greater rate of expansion than any of the hyperscale vendors.

Can you give us a sense of the take-up rate of OCI?

Chelliah: Let’s just say the market is big. Just in Asia-Pacific, I think IDC put the five-year compound annual growth rate at about 34%. In the next three years or so, it’s going to be about a $76bn market in Asia-Pacific alone.

There’s a tremendous amount of latent demand where customers who would like to go to the cloud, but can’t do so for whatever reason, are looking for a path to get there. We’ve come in with a very viable solution with OCI and made it available at customer premises and as a hyperscale cloud service in multiple regions.

In Japan, where we launched OCI more than six months ago, we saw one of the fastest adoption patterns for any of our new regions. And in every region since then, we launched and sold out. It’s just that latent demand, whether it’s in Mumbai or Seoul.

Take the Indian telco Jio, for example. They have hundreds of millions of mobile subscribers and were running on our competitor’s IaaS (infrastructure as a service) platform. They tried us with a small segment of their customer base, and they had an order of magnitude performance improvement. As a result of that, we’re now in conversations about other things.

But my point here is customers see the tangible performance difference of not having to pay the hypervisor tax. We’ve also done some great optimisations at the networking tier as well.

You mean the software-defined networking aspects?

Chelliah: Yes, what we’ve done is off-box networking, where we pushed the whole networking stack to a different box. That means we can guarantee the number of hops between two environments, and our competitors can’t do that. The customer sees that through our performance SLAs (service-level agreements) while our competitors only provide availability SLAs.

And with our second-generation cloud, because of the way we’ve designed our platform, customers can consume any amount of CPU, memory and storage they need, without being limited to the resource ratios set by other cloud services.

What kinds of workloads are your customers running on OCI?

Chelliah: We have a cool mix. While most cloud success stories are about new development projects, we have a whole bunch of customers that are picking up existing enterprise class workloads and running them on OCI.

For example, JNE, an Indonesian logistics firm, is moving all their systems, from real-time logistics tracking to warehouse management, to OCI, while Thailand’s Metropolitan Electricity Authority, is looking to move all their on-premise workloads to the public cloud. That’s the real litmus test, because spinning up a virtual machine and building tomorrow’s application can be done on any cloud, not just ours.

Are they mostly Oracle workloads do they include applications from your business software competitors?

Chelliah: A lot of them are Oracle applications that use the Oracle database. But they could also be third-party or SAP applications that run on the Oracle database, because the hardest problem has been to get a secure, high performance, high throughput, database environment on any of the cloud vendors. That has been the roadblock. Once we move the data, we take that problem away and customers are more than happy to bring the rest of the application to OCI.

And when we start showing customers what they can do with the Autonomous Database in the cloud, they start to apply machine learning algorithms against that database. While we start with the database, it’s not just taking your database from on-premises to the cloud. We show customers that by having data on the cloud, they can benefit from machine learning and AI (artificial intelligence).