When we write the history of the IT era, the big factor that has played an important but not well recognized role will be hardware — specifically, the Oracle Exadata appliance that puts databases into memory for fast access and updates.
All of the cloud software starting with the autonomous database on view this week at OpenWorld — like security, self-patching capability and apps — would be vastly different if the database was still running primarily on disk drives. In fact, the database would be running so slowly (relatively) that some of these solutions would remain theoretical.
But Exadata lives. It is integral to everything Oracle does and everything it announced at OpenWorld 2018 this week in San Francisco. It is such a differentiator that the rest of the database vendor-scape is in serious danger of becoming irrelevant.
After decades of staid, steady progress, the relational database market might be returning to its Wild West roots. The reason is Oracle’s recently deployed autonomous database.
To say it’s just another release of a decades-old product would be to miss quite a bit. Oracle’s upturn relies on the convergence of hardware, software, artificial intelligence and machine learning, and a sharp turn to cloud computing. The only thing preventing a reprise of the 1980s database wars is a lack of competition.
You could have seen all of this coming a few years ago if you knew what you were looking at and had the vision of someone like Larry Ellison. It began when Oracle introduced the first Exadata machine that could keep most, and eventually all, of a business’ database in memory, which greatly accelerated database performance.
In-memory databases operate at nanosecond speeds, while disk drives — the traditional home of databases — run at millisecond speeds. In-memory data operations can run about a million times faster. Although all of that performance doesn’t reach a user just wanting to get some work done, enough does to make in-memory databases revolutionary.
The second innovation announced at last year’s OpenWorld, and fully released earlier this year, is Oracle’s autonomous database software. The product uses artificial intelligence and machine learning to do several things, including installation and tuning, as well as maintenance.
The autonomous database can patch itself without human help. It also can track down intrusions and attempts at data theft that bedevil modern IT departments. Often a patch to thwart intrusions exists, but human operators historically have needed to schedule downtime to implement patches. That’s all over now. However, the full benefits come only when you combine hardware and software.
Third, there’s cloud computing. As you might imagine, buying and installing all of the hardware needed to support the autonomous databases is a big undertaking that is affordable only for big IT shops. That’s why cloud computing is so important. The cloud makes all of this affordable for virtually any business that wants to put its IT in the cloud. That’s good for Oracle too, since it gives the company a very large market to address and little competition.
A Tale of Creative Destruction
At Monday’s keynote, CTO and founder Larry Ellison announced that the company would make the same cloud data center technology available to very large and well-heeled companies that don’t want to share cloud technology. This effectively makes all of the autonomous and in-memory technology available to anyone who wants it.
It’s hard to see how these developments won’t significantly change the database market. Oracle has chosen Amazon Web Services and its databases for transaction processing and data warehousing as its foil, and in demonstrations on stage running one benchmark after another, Oracle’s configuration outperformed the competition by factors of 8 to 1 or more.
Even Oracle’s database running on Amazon without in-memory assist runs significantly slower than the Oracle Cloud. To top off the demonstrations, Ellison showed that his company’s products are less expensive to use too.
All this calls into question why a business seeking cloud IT services would seek them from another party. It also opens up a gap between Oracle and the rest of the database market. The combination of hardware and software that Oracle has produced, along with the long lead times and high costs needed to duplicate Oracle’s feat, make it hard to see how any competitor could bring a credible alternative to market for possibly years.
Taken together, all of this is more than a technology story. It’s a tale of economics — specifically of creative destruction. Advances in technology have begun to commoditize the tech industry, and Oracle is trying to accelerate this curve.
Cloud computing, especially, is a form of commoditization in which basic compute services can be delivered for a fraction of the costs usually involved in supporting all of one’s IT needs in-house.
Markets trend toward monopoly or possibly oligopoly. Oracle has gotten to a rarified position ahead of its competition, in part because it has pursued a program of creative destruction on itself. For too long the company appeared to be making little progress in cloud computing, but in reality it had a swarm of customer-partners, including Salesforce. Engineers worked on site to harden its conventional database products to support cloud computing.
The things the company learned were essential to developing the products it offers today. For instance, Oracle introduced its first Exadata product about one year after it bought Sun Microsystems, and it began selling those units to the cloud industry, at US$1 million per, immediately.
What we’re seeing this week at OpenWorld is the distillation of a great deal of experience in the real world. The result is a new IT paradigm that’s hard to compete against. This technology is Oracle’s hole card. Other vendors might have more cloud data centers, but Oracle is demonstrating that it is the end of an old paradigm, not the beginning of something new.