Larry Ellison Gets Serious About Multi-Cloud. What Is Oracle's Next Move?
At a pivotal time for its database strategy, clear answers are not yet forthcoming.
Larry Ellison is betting Oracle’s database future on a newly articulated multi-cloud strategy. It’s a smart move, albeit long overdue.
But make no mistake—this is an important advance for Oracle and its customers. No longer are Oracle databases confined largely to its own Oracle Cloud Infrastructure (OCI).
Ellison has been talking up multi-cloud for the past few months.
In April, he revealed plans to make Oracle’s MySQL HeatWave database available on AWS and Microsoft Azure (see ‘Oracle vs. AWS’ below).
In July, he unveiled Oracle Database Services for Microsoft Azure.
In September, he announced availability of MySQL HeatWave on AWS.
Next: MySQL HeatWave will be available on Microsoft Azure “in the near future,” Ellison said.
“The multi-cloud era has begun,” Ellison declared during Oracle’s Q1 FY23 earnings call on Sept. 12.
I would argue that the multi-cloud era actually began a few years ago, and that Oracle is now catching up, but better late than never. It’s exactly what Oracle needed to do amid growing competition in the cloud database market—both from the Big 3 hyperscalers (AWS, Google Cloud, Microsoft Azure) and from fast-growing companies like Cockroach Labs, MongoDB, Snowflake, SingleStore, and others.
The fact that Ellison, Oracle’s CTO and spiritual leader, has thrown his endorsement behind multi-cloud is a positive development for Oracle’s future direction. The big open question is, What’s next? What additional steps will Oracle take to make its database technology available in other clouds, as well as to make it easier to run non-Oracle databases on OCI?
The limits of Oracle-on-Oracle
Without a better multi-cloud story, Oracle was at risk of becoming too narrowly typecast as an Oracle-on-Oracle cloud database provider.
“I know a lot of people for years have been concerned about, Can Oracle sustain its leading marketshare in the database business?” Ellison said. “And I think what is clear is if our databases are available in multiple clouds, I think then the answer is clearly yes. And if our database is not available in multiple clouds, then it’s an interesting question, whether we can maintain it just in our own cloud.”
As I detailed in an earlier blog post (‘Oracle vs. AWS’ below), an all-Oracle stack—Oracle database on Oracle cloud on (in some cases) Oracle hardware—may be too much Oracle for some customers in today’s multi-cloud world.
The verdict is in on the question that Ellison shared regarding whether Oracle can maintain database marketshare leadership exclusively on OCI—and the answer is clearly “no.” As I have covered before, Oracle’s database marketshare has declined over the past two years, according to Gartner.
In 2020, Oracle dropped into second place behind Microsoft, and in 2021, Oracle fell into third place behind #1 Microsoft and #2 AWS, according to Gartner. The chart below shows Oracle’s gradual marketshare decline.
So it’s no coincidence that Oracle has targeted Microsoft and AWS with recent multi-cloud database announcements, as together they account for 47.9% of the market compared to Oracle’s 20.6%.
Oracle is hoping to lure database users away from AWS by making it easier for them to: A) choose its MySQL HeatWave or newly developed apps on AWS, or B) migrate existing AWS databases to MySQL HeatWave without leaving the AWS cloud (and thereby avoiding data-egress fees).
Here’s how Ellison explained it back in March, when he first revealed plans to make MySQL HeatWave available outside of OCI:
“We're going after the Aurora user base and the Redshift and Snowflake user base,” Ellison said. “We want to make it really easy to convert from Aurora and Redshift, or Aurora and Snowflake, to Oracle HeatWave. And if we’re running an AWS, for example, you press a button, a couple of buttons, and your data is moved immediately to Oracle MySQL HeatWave. You do not have to change your application at all.”
On the more recent Sept. 12 earnings call, Ellison reiterated Oracle’s goal of winning AWS customers.
“I personally have been talking to some of Amazon's most famous brands that are running at AWS,” he said. “And the AWS build is getting very large, and they can save a huge amount of money by moving to OCI. And we—I expect next quarter—will be announcing some brands, some companies moving off of Amazon to OCI that will shock you.”
Oracle’s ‘best and greatest’
This aggressive approach—unapologetically targeting AWS users on AWW’s own turf—is very different from the July announcement of Oracle Database Service for Microsoft Azure. At the time, Ellison and Microsoft CEO Satya Nadella jointly appeared in a video to share the good news.
So, there are different ways to coexist in multi-cloud. You can play nice, as Oracle is doing with Microsoft. Or you can throw elbows, as Oracle is doing with AWS. Both can work. The key thing is that customers have more options and flexibility, as Ellison emphasized.
“We’ve decided to make our best and greatest technology available in multiple clouds,” Ellison said. “And that gives customers choice.”
Oracle gets credit for what is clearly a step in the right direction with its expanding multi-cloud offerings. That said, there are some missing pieces.
What about Google Cloud?
As Editor of the Cloud Database Report, one of the first things I want to know about any database provider is what clouds they run on. The answer I’m looking for is AWS, Microsoft Azure, and Google Cloud—all three—because that is really the ultimate in customer choice.
But Oracle’s database offerings across those cloud platforms are a mixed bag. MySQL HeatWave is not yet available on Azure, though Oracle says it’s coming soon. And there’s been no mention whatsoever of MySQL HeatWave on Google Cloud.
Oracle seems to have a blind spot when it comes to Google Cloud. As far as I can tell, Ellison hasn’t mentioned Google Cloud is his recent comments about multi-cloud. But Google Cloud is rising in the cloud database market, and last year moved into the top five for the first time, according to Gartner.
It is possible to run Oracle workloads on Google Cloud. Here are some of the ways to do that, as shared with me by Google Cloud:
“Google today offers low-latency connectivity between Oracle Databases and Google Cloud via the Bare Metal Solution. Google can provide latency to OCI similar to the Azure/OCI interconnect via Equinix in available regions. Additionally, Google offers a broad breadth and complete suite of databases such as Cloud SQL, Spanner, Bigtable, AlloyDB and more. We also have database migration offerings that give customers the ability to run a variety of workloads on GCP.
Note that Google Cloud highlighted its database migration services—which are specifically targeting Oracle workloads, among others. All of the cloud database vendors have migration tools and services, and its an area of continued development and activity.
In other words, it’s not enough to run and manage third-party databases in a multi-cloud environment; you must be able to move the data from one to the other, as well.
Oracle responds
What’s next in Oracle’s multi-cloud strategy? I asked Oracle a few specific questions about that, and its answers were opaque and uninspiring.
My question: Are there any plans to make MySQL HeatWave available on Google Cloud? And if not, why not?
Oracle’s answer: “MySQL HeatWave is a core part of Oracle’s Distributed Cloud strategy and is available in multiple clouds, including OCI, AWS, and Microsoft Azure in the near future. When we have additional information, we will share that with you.”
My question: Are there any plans to introduce Oracle Database Services for AWS or Google Cloud, similar to what Oracle is doing with Microsoft Azure?
Oracle’s answer: “Since 2019, Oracle and Microsoft have partnered to deliver the high speed interconnect between Microsoft Azure and OCI. MySQL HeatWave runs natively in the AWS cloud. It’s a true native experience for AWS customers, including the data plane, the control plane, and the management console.”
Oracle must be able to provide more insightful answers to these and similar questions. I don’t see the slightest hint of plans by Oracle to do more in those areas, which is why I’m not more bullish about Oracle’s overall game plan.
I would think that CloudWorld in October would present an ideal opportunity for Oracle to articulate its next steps in multi-cloud with greater clarity.
MongoDB, PostgreSQL, Redis, Cockroach, et al.
Another way to look at multi-cloud is the availability and ease-of-deployment of third-party databases and database services on AWS, Google Cloud, Microsoft Azure, or Oracle OCI.
Google Cloud, for example, supports MongoDB, Neo4j, and Redis. And Microsoft offers Postgres, MySQL, MariaDB, and Apache Cassandra compatibility in the form of Azure services.
How does Oracle compare? It is possible to run non-Oracle databases on OCI. Oracle provided the following resources on ways to do that:
Here’s a webpage with more info and resources on Oracle’s multi-cloud offerings.
Also, earlier this year, Oracle announced a MongoDB API for the Oracle Autonomous Database. See my blog post below for details on that.
Laggard or leader?
All of the above begs the question: Has Oracle become a leader in multi-cloud databases?
Until recently, I considered Oracle a laggard, but its latest advances are cause for reassessment. However, I’m holding back on calling Oracle a market leader in multi-cloud until I hear straight answers to my questions about the holes that remain.
The bottom line is that Oracle is showing good progress toward becoming a bonafide multi-cloud database provider, in part because Ellison deemed it a priority, but also because market pressures were forcing Oracle in this direction.