Oracle, Google Cloud, Cockroach Labs, Yugabyte, and the Book of Kells
The database competitive landscape is being upended by multi-cloud, GenAI, and now more than 1,000 platforms.
Welcome to the Cloud Database Report. I’m John Foley, a long-time tech journalist, including 18 years at InformationWeek, who worked in strategic comms at Oracle, IBM, and MongoDB. I invite you to subscribe, share, comment, and connect with me on LinkedIn.
Let’s catch up. I’ve been traveling a lot from my home base in New York — with trips to Boston, Chicago, Denver, Las Vegas, San Francisco, and Florida.
Next stop is Dublin, Ireland, to visit my daughter, a grad student at Trinity College. We will visit the Book of Kells, the 1,200-year-old, 680-page calligraphed manuscript known for its magnificent, intricate illustrations inked onto calfskin folios.
It will be interesting to contemplate this medieval “content” amid the rise of AI-powered image generators. Two thoughts on that: I have no doubt that the priceless, physical manuscript will be a visual experience with no parallel in AI. On the other hand, I can also imagine the Book of Kells and the story it tells being brought to life in brilliant new ways by GenAI. The Book of Kells was recently digitized, so it’s probably only a matter of time before that happens.
I asked Anthropic’s Claude LLM if it had been trained on the Book of Kells. Claude responded that it doesn’t have direct access to the Book and advised me to go to the Trinity College Dublin Library, which is exactly what I plan to do.
Here’s my take on what’s been happening in the database market.
Oracle hits a turning point
It’s time to begin talking about the new Oracle. With blockbuster multi-cloud database agreements with Microsoft, Google Cloud, and AWS (in that order), Larry Ellison has flipped the script on the old trope that Oracle is not to be tangled with. It’s now the picture of diplomacy.
And Oracle didn’t merely make “announcements” of partnerships to colocate its databases and Oracle Cloud Infrastructure (OCI) in the data centers of the world’s three biggest cloud providers. Larry Ellison had very public one-on-one conversations with Microsoft CEO Satya Nadella, AWS CEO Matt Garman, and Alphabet/Google CEO Sundar Pichai. That leadership engagement underscores the strategic nature of this next level “coopetition” for both Oracle and its new best friends.
After attending Oracle CloudWorld in Las Vegas in early September, I’m optimistic about what this means for Oracle’s enterprise customers. They will now find it easier and better to run Oracle databases on non-Oracle clouds. Not only will performance be on par with what they get on OCI and at similar cost, but there’s less complexity than with the Do It Yourself (DIY) model.
The new Oracle Database@ family of services are a turning point in a market that had grown so accustomed to competitive head-butting that many were surprised when Oracle and AWS put aside their differences and AWS CEO Garman stepped onto the CloudWorld stage with Ellison.
Oracle is betting its financial future on the success of this strategy. In its recently reported Q1 FY25, cloud database services revenue grew 23% to $2.1 billion. “We expect those cloud database revenues collectively will be the third leg of revenue growth alongside OCI and strategic SaaS,” CEO Safra Catz said on the earnings call.
Google Cloud’s three-pronged GenAI strategy
Speaking of Oracle CloudWorld, I ran into Andi Gutmans, VP and GM of Databases for Google Cloud, who was speaking in a session on multi-cloud with Oracle and Accenture. Gutmans described a “fully integrated experience” with Oracle Database@Google Cloud and a “unified vision” for using Google technologies (Vertex AI, Gemini, BigQuery, Looker, and even Google Maps) with Oracle databases.
As I said above, the competitive landscape has changed. Mountain View and Redwood Shores are now closer than ever.
Gutmans and the rest of the Google Cloud database team continue to push out tech upgrades faster than I can keep up with. In a recent blog post, Gutmans outlined three areas of focus with GenAI:
“First, helping developers build intelligent apps with operational data. Second, leveraging genAI — and specifically our Gemini models — to simplify every stage of the database journey, including migrations, fleet management, troubleshooting, and performance optimization. And third, modernizing your databases so you can take advantage of gen AI.”
Here are some recent developments:
The ScaNN vector index for AlloyDB — a fast, PostgreSQL-compatible index that can scale to more than a billion vectors — is now GA.
A new multi-cloud managed cloud service from Aiven can be used to deploy, manage, and scale AlloyDB Omni on Google Cloud, AWS, or Azure.
New capabilities for Memorystore — Google Cloud’s in-memory data store — include Memorystore for Valkey 7.2 (an open source key-value store), plus vector search for Redis Cluster and Memorystore for Valkey 7.2 services.
Firebase Data Connect is a new “backend-as-a-service” (in preview) that is integrated with a fully managed PostgreSQL database powered by Cloud SQL.
Introduction of Database Center, an AI-assisted dashboard, to monitor “fleets” of Google Cloud databases through a single interface across projects and regions.
Also: SQL Server migrations to Cloud SQL for SQL Server are now GA as part of Google Cloud’s Database Migration Service.
There’s more in Google Cloud’s Database News Roundup here, including a case study of Google’s migration of billions of Gmail users to its Spanner database, which it describes as one of the largest data migrations ever.
Cockroach Labs and the resiliency experience
I attended my third annual Cockroach Lab’s RoachFest conference in New York City in late September. The theme was the need for businesses to shift from a disaster-recovery mindset to “operational resiliency,” which isn’t exactly a new idea, but one that businesses are obviously having a hard time realizing.
Less than one week after RoachFest, the Bank of America became the latest big enterprise to suffer a far-reaching system failure affecting thousands of its customers. In some cases, accounts showed zero balances. Very disconcerting for anyone to have their money disappear like that.
Such miscues continue to be all too familiar. The CrowdStrike incident in July was described as at the largest IT failure ever. And last year, I documented a meltdown of the FAA’s NOTAM system, which resulted in air traffic delays across the US.
In my experience, one of the primary issues causing computer and network failures is complexity, which leads to fragility. Legacy tech, as in the case of the FAA’s NOTAM, makes things worse. And there’s human error. A lot can and does go wrong.
There’s a sliding scale when it comes to business readiness for these glitches and gotchas. Cockroach CEO Spencer Kimball showed how organizations can go from 99% system availability on individual machines, to 99.9% in data centers, to 99.99% across regions, and 99.999% in the cloud.
The big idea for IT is to architect for a “resiliency experience.” In other words, the business and its customers are unfazed by short-lived outages because they are self-healing and low impact. We need this in today’s interconnected world because a single little hiccup like CrowdStrike’s content configuration update can ripple across billions of systems and devices.
“Anything that can go wrong, does go wrong,” Kimball told me. “You simply have to assume a constant rate of breakage, and then design extremely redundant systems.”
Cockroach Labs, with its “survivable” database architecture, was created to meet this challenge. And the company’s customers are making progress, but there’s still a long way to go across the business landscape.
Kimball envisions a future where CIOs and CTOs can build and manage data systems that are “above the clouds.” What does that mean? Two years ago he described a future-state “infrastructureless” architecture, with CockroachDB as a global service. Last year, our talked about regionless databases. Both of which are somewhat abstract concepts, but the vision is compelling. The way to do it, Kimball says, is to separate the control plan from the data plane, so data and availability can be managed across clouds, data centers, and global regions — a.k.a. above the clouds.
Those are my takeaways from RoachFest 2024. Also, there are two recent developments to mention. First, Cockroach is changing its database licensing scheme, effective in November. It’s eliminating its Core offering for self-hosted DBs and creating a two-tier Enterprise license with a paid tier for companies with $10 million in annual revenue and a free tier for smaller orgs. You can see the details here.
And Cockroach has introduced vector search in the latest release of Cockroach DB, release 24.2, which is in preview.
Yugabyte wants its Postgres cake
Yugabyte in mid-September announced that it described as significant architectural changes for enhanced Postgres compatibility to its database. There are two key innovations.
First, an Adaptive Cost-Based Optimizer has been extended to better support high-scale and multi-region apps. The Optimizer determines a query plan based on whether data is co-located, automatically sharded, or distributed across zones or regions.
And second, Smart Data Distribution determines whether to store tables together for lower latency, or shard and distribute them for scale. This new feature enables YugabyteDB to support over 50,000 objects (tables, indexes, etc.) versus 2,000 before.
Smart Data Distribution reuses the PostgreSQL query engine to achieve runtime compatibility with PostgreSQL. In fact, more than 92% of PostgreSQL query patterns now run with comparable performance on YugabyteDB, up from 47% two years ago.
What do these changes amount to? Yugabyte wants to have its Postgres cake and eat it too. In other words, customers can now use YugabyteDB to run even more Postgres-compatible workloads, which is essential given the popularity of Postgres. And they can also build and deploy the kind of distributed workloads that are typical of enterprise IT environments.
For more, see this explanatory blog post by CEO Karthik Ranganathan and VP Suda Srinivasan, both of whom I talked to about these new developments. Yugabyte will go even deeper into these new features and its future direction at the company’s Distributed SQL Summit 2024 on Nov. 12 in Salt Lake City, which I hope to attend.
Other news…
Mike Stonebraker’s new startup, DBOS (i.e. database operating system), has hired Jeremy Edberg as CEO. In this Q&A, Edberg discloses that he has received three Emmy Awards and is a world record holder for gift giving. His vitae includes eBay, Reddit, Neflix, Amazon, and Lambda.
Tencent, the Chinese hyperscaler, has developed a multi-modal, NoSQL database, according to The Register.
Carnegie Mellon University’s Database of Databases has surpassed 1,000 DBMSs. I predicted in January that this was going to happen, and now the milestone has been reached.