We have bigger targets than beating Oracle, say open source DB pioneers
Advocates for MySQL and PostgreSQL see broader future for movement they helped create
MySQL pioneer Peter Zaitsev, an early employee of MySQL AB under the original open source database author Michael "Monty" Widenius, once found it easy to identify the enemy.
"In the early days of MySQL AB, we were there to get Oracle's ass. Our CEO Mårten Mickos was always telling us how we were going to get out there and replace all those Oracle database installations," Zaitsev told The Register.
Speaking at Percona Live, the open source database event hosted by the services company Zaitsev founded in 2006 and runs as chief exec, he said that situation had changed since Oracle ended up owning MySQL in 2010. This was as a consequence of its acquisition that year of Sun Microsystems, which had bought MySQL AB just two years earlier.
As an alternative relational database to Oracle, MySQL had won early fans among the second generation of online companies, including Facebook, Google, and Uber. While advocates had been horrified at the prospects of the world's largest proprietary database vendor getting its hands of their open source baby, Big Red's stewardship of the system has not turned out exactly as feared.
"They were thinking, 'Oh no, Oracle will kill MySQL, make it completely proprietary' and so on. You can argue about their motives, but Oracle has been doing pretty good engineering in MySQL," Zaitsev said.
But it is qualified praise, which marks out the nuances developing in the open source database market.
The co-author of High Performance MySQL: Optimization, Backups, and Replication pointed out that "if you're an end user, MySQL 8 is a great database" with a lot of good features and performance.
What about the developers?
However, if you are a developer or a packager of other systems requiring access to the MySQL source code, "Things are not quite so rosy," he said. The reason is Oracle offers a "drop-ship" approach to open source development. While Oracle releases source code, it does not offer an open discussion with its developers so they can get feedback before features go on general release.
"They don't do development in public," Zaitsev said.
The intent may be to make life harder for other businesses trying to make money on the back of MySQL, such as AWS with its Aurora MySQL compatible system, or MariaDB, with its MySQL fork, but it also creates barriers for engineers who want to look at the code in development to see if there might be easier alternatives to a monolithic Oracle download.
To be open or not to be open
Another trend muddying the open source waters alongside Oracle's "drop-ship" development is the so-called single vendor open source systems, whose companies back a separate foundation running the open source project.
While the companies MariaDB and MongoDB maintain their databases are open source, there are important differences from "true" open source software for users who become wedded to the cloud services they offer, according to Zaitsev.
"First of all, the [MariaDB] foundation gets a lot of funding from MariaDB corporation, and a lot of the development is done within the corporation, so it's not that easy," he said.
Meanwhile, the complete solution may rely on proprietary software such as the MaxScale, which extends high availability. At the same time, only customers get direct access to the MariaDB code. Even though they can republish the code, the move still puts non-customers at a disadvantage, Zaitsev claims.
Similar arguments apply for MongoDB's Atlas DBaaS.
"If you don't have a commercial relationship, you essentially lose access threats to that source code," he said.
Since founding Percona, Zaitsev has helped build up the services business to around 350 employees, across over 50 countries funded only by its own revenues. As well as providing services and consultancy backing open-source database systems, it also offers open-source software tools, with around 75 million downloads annually.
The software it offers now includes a SaaS database management platform initially targeted at PostgreSQL, MySQL, and MongoDB.
But the software itself remains open source, even if customers end the commercial relationship with Percona, Zaitsev said. "If you choose not to be our customer, you have access to all our software. You can hire somebody else to maintain it," he said.
Meanwhile, the pitching of open source against Oracle's own proprietary database has shifted as the market has moved on and developers lead a database strategy building a wide range of applications in the cloud, rather than a narrower set of business applications.
Zaitsev pointed out that if you look at the rankings on DB-Engines, which combines mentions, job ads and social media data, Oracle is always the top RDBMS.
But a Stack Overflow survey would not even put Oracle in the top five. So as developers are concerned, the debate about whether Oracle is the enemy is over. "The reality is, the majority of developers — especially good developers — prefer open source," he said.
It is not just Percona making these arguments about the purity (or lack thereof) of open source database software.
The EDB perspective
EDB, which supports and contributes to PostgreSQL, makes similar points. It too started out as an Oracle killer but is also moving on.
Bruce Momjian, EDB vice president and core member of the PostgreSQL development group since 1996, told The Register: "When I started with EDB in 2006, we really pitched ourselves as the Oracle alternative. And, you know, that was a good run for maybe 10 years or 12 years."
But what has changed in the past five years, is the market for customers who either want to stay with Oracle, or an Oracle-compatible RDBMS built on PostgreSQL, is shrinking.
"There's a lot of companies now who are basically saying, 'Forget the Oracle API, I want to standardise on the PostgreSQL API.' They don't even want a non-PostgreSQL API because they see it is a growing market and opportunity with additional cost savings, flexibility, and continual innovation," he said, also speaking at Percona Live.
"Years ago, if you had to rewrite your application from Oracle to PostgreSQL, that was a negative, that was a cost to you. Now, if you're writing to work on a PostgreSQL, they think that's a value-add to that organization. Leaving the Oracle API is no longer seen by everyone as a negative, it's actually a positive to some people, and that helps to justify the cost," Momjian said.
PostgreSQL is available under the OSI approved PostgreSQL Licence and has such wide developer support it is difficult for single-vendor open source databases to keep up, Momjian said.
He cited Greenplum, which used PostgreSQL as the basis of its data warehousing system, which he said limited success. "They were going to be so good, they said that, 'We don't need to stay with what the community is doing'."
"But time proved them wrong. It's very hard for a company to innovate at the speed of PostgreSQL. Oracle can't do it. If Oracle can't innovate at the speed of Postgres ... How do any of these other people think they can do it?"
Developers want more than open source
Oracle would of course beg to differ. Last year, it beefed up Exadata X9M, and claimed it achieved online transaction processing (OLTP) with more than 70 per cent higher input/output operations per second (IOPS) than its earlier release, the X8M. Building on the hardware integration it inherited via the Sun Microsystem buy – and which is only available in the Oracle cloud or on-prem systems – Big Red also said the system performs with 19µs I/O latency from database to storage, 10 times faster than flash memory.
For retention, not acquisition?
Despite the performance, its primary target is existing customers, Carl Olofson, IDC research vice president, said at the time. Even if Oracle is no longer the main opposition, the story is not quite over yet. Speaking to us again, Olofson argued it is too simple to say developers are only looking for open source databases on which to build their new projects.
"Up until recently the selling proposition for open source was that, even with a package of enterprise features build around it, it was still cheaper than entirely proprietary package from Oracle or Microsoft. But now that we're going into the cloud it all gets murkier," he said.
Since customers are paying service fees to support enterprise grade open source RDBMSs in any case, when companies offer a database as a service in the cloud, the charging model is the subscription model, which customers are comfortable with, and do not necessarily care whether what is behind it is technically open source or not, he said.
"At the end of the day, most customers just want value for money. They want to know if they're paying a reasonable price and they're getting a level of service and performance that is suitable for what they're trying to do. A lot of surveys show preference for open source, but we found that when you peel it back, it's really a preference for open source because they see the interface as a standard, and that means they can hire people with MySQL experience or PostgreSQL experience fairly easily," Olofson added.
Whether they would prefer open source to proprietary so they could eventually port the code should they want to is a moot point, he said. Meanwhile, Olofson is not convinced that open source can always out-innovate proprietary systems either, especially when a new generation of innovators are coming onto the markets.
"There are some excellent technologies coming onto the market — the geo-distributed CockroachDB system with transaction support, for example. Asked about going open source, they say, 'Why would we do that? This was hard to develop and our investors expect to return on investment'."
It makes sense when the technology is mostly known which is the case of PostgreSQL and MySQL. They have these massive QA teams, they're certainly battle-tested, and for most common use cases, they're perfectly fine. But when you get into some of these more extreme use cases the question would be where did the open source databases stand?
"What we call extreme today is becoming more common because of the increase in data volume and the increase in use of streaming data, and other technologies that really stress data systems and require new and different kinds of technology to handle," Olofson said.
For those believing in the open source ethos, Oracle may no longer be their main enemy. But the battle may be just beginning. ®