Oracle is slipping in its hold on the database market, but it's going to take a long time to topple.
Oracle's annual revenue has been flat for the last five years, with new license revenue in steady decline during that same period. According to co-CEO Mark Hurd, however, Oracle can decline for decades and still retain its top spot on the database heap. He may be right.
Despite growth from would-be challengers, the database provides the ultimate lock-in. As such, the real question is not how fast companies will dump Oracle for rival databases, but rather how fast cloud computing will grow, given Oracle's relative weakness in that area.
Using DB-Engines, I dug into the popularity trends for a variety of databases. DB-Engines uses an amalgam of job postings, Google searches, Stack Overflow mentions, and more to calculate popularity. Importantly, the company measures relative popularity.
SEE: Cloud database growth may be slowing as lock-in fears loom (TechRepublic)
Looking at the top-10 databases over the last year, the big winners have been Microsoft's SQL Server and the open source PostgreSQL, with Oracle and MySQL both in decline:
Looking out over a longer period of time at trends, Oracle's relative decline in popularity is pronounced:
Microsoft's SQL Server had been in decline but has recovered somewhat in recent years:
Interestingly, though it had stumbled of late, MySQL has been growing in this similar period of time:
PostgreSQL, by contrast, has been growing from strength to strength:
As has MongoDB, recently reported to be exploring an IPO this year as it climbs above $100 million in revenue:
If you're an Oracle hater, it's not time to pop the champagne. Despite these trends over the last few years, Oracle still dominates the database landscape and will for a long time.
Trading lock-in for lock-in
Though cloud databases like Amazon Aurora have been on a tear, Hurd projects Oracle will lose roughly zero market share to them:
How much database market will Oracle lose to [Amazon] Aurora? My guess is close to zero. The third largest database in the world is IBM DB2, and it's been going out of business for 20 years. If it was so easy to replace databases, DB2 market share would be zero.
In short, lock-in may not be a CIO's friend, but it's clearly Oracle's. And IBM's. And Microsoft's.
Not that these companies are holed up in some executive board room somewhere, nefariously laughing at evils they're inflicting on the world. Well, they might be, but the simple truth is that, as Hurd argues, it's hard to rip-and-replace a database. Once the data goes in, it's bothersome to get it out. This is why lock-in shy CIOs should think twice about where they're shoveling their data even as they race to put more workloads in the cloud.
By some measures, they are.
SEE: Want to avoid cloud lock-in? It's about the database (TechRepublic)
There are indications that cloud databases (e.g., Amazon Redshift, Google Cloud SQL, Microsoft Azure DocumentDB, etc.) are slowing down in terms of growth. While hard data is hard to come by (e.g., I've heard that Amazon Redshift's growth has slowed, which certainly shows up in a DB-Engines' trend chart, but haven't been able to confirm this with absolute data), it makes some intuitive sense if we think about why enterprises might be reluctant to lock their data in with a cloud vendor (or any vendor).
As a friend who has spent time at several major database vendors put it to me, "Cloud lock-in and s____ products are making big buyers wince." Asked to explain, he went on:
Companies that are born in the cloud are ok with [the prospect of cloud database lock-in], whereas big companies who lived through the hardware vendor lock-in days or stored procedure or ERP lock-in are all weary of it and see it unfolding all over again. It's a generational thing.
Speaking of "generational things," another database veteran familiar with Google's operations had this to say: "At Google Cloud they couldn't get any cloud database growth out of Postgres. Their customers all wanted MySQL, but Google couldn't get engineers to work on MySQL because they viewed it as boring."
So on the one hand, we may have enterprises loathe to get locked into yet another big vendor, wanting to keep dropping data into tried-and-true MySQL, while the cloud vendors want to move on to shiny, new...and deeply proprietary databases. It's a match made in heaven. Or hell.
SEE: NoSQL keeps rising, but relational databases still dominate big data (TechRepublic)
None of which is stopping cloud adoption. As Gartner has been pointing out for years, public cloud VMs have been on a torrid 20X growth trajectory even as private cloud VM growth has proceeded at a more comfortable 3X growth rate. Lock-in or not, the cloud keeps booming. There are signs, however, that CIOs learned their lesson from their last round of lock-in with Oracle et al.
Mark Hurd can take comfort that database lock-in will keep the lights on at Oracle for years to come, but most of the growth in the database market seems to be happening in open source (cloudy or otherwise). That trend is good for CIOs grown weary of lock-in. It's not so good for Hurd's paycheck.
- 2016 was a bad year for NoSQL databases, and the cloud is to blame (TechRepublic)
- Special report: The cloud v. data center decision (free PDF) (TechRepublic)
- Google adds the last piece to its cloud database stack (ZDNet)
- Which cloud will give you the biggest bang for the buck? (ZDNet)
- Cloud computing policy (Tech Pro Research)