This article is more than 1 year old

Larry Ellison fought internal battle to kill Oracle's first-generation cloud

But Gen 2 faces an uphill battle in market share and technology

Oracle founder and CTO Larry Ellison canceled its first cloud after an internal struggle, he told investors gathering at the company's CloudWorld event.

At the conference in Las Vegas last week, Oracle announced several new distributed cloud services including Oracle Alloy, a way for customers to rebadge Oracle's Cloud@Customer service to offer their own. It also announced that the Oracle MySQL HeatWave database service would be available on Microsoft Azure, and plans to open new public cloud regions in the US, Serbia, and Mexico.

But Ellison admitted these developments were only possible after something of a false start.

"When we [first] were doing our infrastructure cloud, I actually canceled the project. There was a pretty big disagreement between me and the powers that be at Oracle," he told investors. "I thought we were just copying what the other guys were doing – which I thought was a really bad idea – and I wanted to start over and do this Gen 2 cloud," he said.

Announced at Oracle OpenWorld in 2018, Oracle's Generation 2 Cloud introduced a new approach to architecture, putting customer code, data, and resources on a bare metal computer, while cloud control code lives on a separate computer with a different architecture. Oracle said the approach meant it could not see customer data while there was no user access to the cloud control code.

Ellison admitted it did not have first mover advantage in the cloud but argued Oracle also came from behind in the database market in the 1990s to become the dominant player. "So how did we ultimately beat IBM? We were faster and cheaper and more secure and more reliable. There's no magic [in the cloud market], we have to be faster and cheaper and more secure and more reliable."

He claimed Oracle was on the brink of signing big brand customers away from AWS. Yet Ellison's bullishness does little to mask the cold reality of the market in which he hopes to succeed.

Oracle may be growing fast, but it is a long way behind the dominant players and others that are growing faster still. A Canalys report from August showed $62.3 billion was lavished on Infrastructure-as-a-Service (IaaS) in the three months ended June, up 33 percent year-on-year.

Meanwhile, AWS, Microsoft, and Google were growing ahead of the market as a whole at 42 percent per annum. Together they accounted for 63 percent of market revenues.

Canalys did not publish figures on Oracle's market share or growth, because it is deemed to sit outside the biggest cloud providers. Oracle said its cloud growth (including SaaS) hit nearly hit 20 percent in Q4. Some research suggests its market share might be as low as 2 per cent for the cloud market as a whole.

Still, the only way is up, Oracle must hope.

In doing so, Oracle acknowledges that the best approach is to admit that most customers run multiple clouds. As such, it has made its MySQL HeatWave analytics service available on AWS, with Microsoft Azure soon to come.

It has also expanded Oracle Interconnect for Microsoft Azure designed to offer organizations a simple migration path to a multi-cloud environment that includes Oracle Database via a low-latency, private connection between the two cloud providers. The interconnect was first announced in 2019.

One commentator attending CloudWorld saw advantages in Oracle's OCI strategy, but also weaknesses.

Speaking to The Register, Keith Townsend, principal analyst at CTO Advisor, said: "What pulls customers to OCI is the network egress: no network egress charge is massively popular. Secondly, colocation with Microsoft Azure. If I want proper cloud services, I can get those from Azure. If I want to host my database, and my static workloads or my systems of record or my enterprise systems in the cloud [you can use OCI].

"It's not talked about, but SAP is a great workload to put on OCI. You can put it in the VMware Cloud solution which is immensely popular because of the low egress charges. That is the strongest argument for Oracle Cloud, that they allow you to bring these enterprise applications to their platform without you changing your operations model."

However, Oracle lagged behind the leading cloud providers in providing a raft of services and applications on top of its cloud infrastructure.

Townsend said: "The biggest gripe [with OCI] is the lack of services. They went from five to 100 services in just a matter of a few years. But AWS now has several hundred services, as does Azure and Google Cloud. So, if you look at cloud-born architectures, and the ability to lean on that expertise in those services, then public cloud providers [can help] build applications faster and cheaper. Oracle simply doesn't have that. They are still basically network, compute, and storage. Oracle hasn't gotten there yet, which can drive up expenses in another direction. So, speed to development is a serious issue for Oracle Cloud infrastructure."

With the technology far from complete, and a sizeable gap between it and the top three cloud providers, Oracle has its work cut out in making an impression on the fast-growing, hundred-billion-dollar cloud market. It may have done it before, but there's no guarantee history will repeat itself. ®

More about


Send us news

Other stories you might like