Big databases life expectancy - 10 years?

Profile picture for user mbanks By Martin Banks March 1, 2015
Summary:
It may be a `minnow’ for now, but MariaDB’s CEO sees big changes coming in the database market as the systems lose their mystique and becomes just another commodity.

Patrik_Sallner_small+picture
Patrik Sallner

Four software vendors – Oracle, Microsoft, IBM and SAP – are reckoned to take the lion’s share of the $30-$40 billion annual revenues generated by selling database systems and their associated maintenance, support and training requirements.

But it won’t remain that way for too long, at least not if Patrik Sallner has his way.

Sallner is the CEO of MariaDB, an eight-year-old contender in the SQL database market. He gives it ten years for the four to see their dominance diminished as the groundswell of cloud, open source and hyperconverged architectures cuts the ground from under them, pushing users in radically a different direction.

At the same time MariaDB should be able to provide a bridge to a fair degree of current database users if only because its history – having sprung out of the original open source upstart, MySQL.

Since it started the company’s database has now been adopted in all the Linux distributions, becoming the default in both Red Hat and Suse. It has also been selected by both Google and Wikipedia.

These big-name reference users have also attracted a growing number of business users, which in turn has prompted the company to introduce MariaDB Enterprise, essentially the same system as used by Red Hat and Suse, which is a subscription-based implementation that includes optimised binaries and a number of tools for monitoring and managing the database.

The cloud alternatives

One of the pressures the four database leaders will face is likely to come from the new cloud service vendors, such as Host Analytics: a company which has targeted users of Oracle subsidiary, Hyperion. Many of these users, like those of other database systems and their associated applications, have avoided updating their applications because they have lacked sufficient cost/benefit for them at a time of austerity. But as the benefits of making the change have come along the costs for multiple intermediate upgrades has been punitive.

But there have been, as Sallner observes, alternative services to turn to in the cloud:

That is absolutely the case and we are seeing it, driven both by the cost issues or the arrival of new applications offering significant advances in performance. In a growing number of instances you would no longer imagine buying applications to run on premise when you look at the alternatives coming available as services. The same applies to mobile apps, which are essentially SaaS. So we are finding that a lot of users are currently evaluating these options.

And we are seeing a move away from buying individual applications and towards going with vendors offering a pre-integrated architecture where the choice is based on what applications components the service providers have available. In this context the database becomes a choice that is no longer made independently but as a part of something that is pre-defined because all the components available work together. Users no longer have to think about the database because it is no longer a critical decision: it becomes commoditised.

He points to Amazon as an example, where it now offers a marketplace of functional components that can be easily pieced together to provide whatever the user needs. And the leading cloud service vendors now provide service building capabilities that can pull together all the components a customer requires to present them with a service package:

We are now working to ensure that MariaDB is available through all those cloud service providers. And none of this is of great interest to the customers, they are not that interested in which database is used, they take the one that is most convenient. By ensuring we are compatible with all the leading SaaS solutions and cloud stacks we can work towards that goal.

Another factor with the cloud that he has observed is the blurring of the edges between where user communities end and cloud ecosystems begin. This is, arguably, the way it should be now. The clear demarcation between technologies and their uses is no longer relevant to the development of agile and dynamically reconfigurable services, even if it is applications which remain the building blocks and bedrock of the services being developed.

Communities of common interest

Sallner says:

Community, in the traditional open source sense, is where a number of developers come together to develop some code, where an eco-system is where a number of different companies ensure their products are compatible with each other. But now there are communities of users as much as a community of contributors, and they are now starting to influence the direction of the services. It is now important to hear their voices, because it is the outcome of the collective voices that is important.

And those communities are starting to build up. He pointed to Salesforce as an example of a vendor with an extremely active user community. Other service providers are following in those footsteps.

MariaDB is already partnering with Rackspace and Pivotal, and is currently in discussion with Amazon, Google and Microsoft to see how they fit into those environments.

The company already has Microsoft-based customers, and as MariaDB has a good degree of compatibility with SQL Server, it is already being used by customers looking to switch existing applications into the cloud. This, he suggests, maps onto what he sees as the natural logic of cloud services being built on open source technologies. It also follows the already established trend of cloud services to be largely built on Linux.

So far the transition to MariaDB has, according Sallner, been pretty straight forward, even when it has had some necessary complications. Wikipedia, for example, was running an almost pure version of MySQL already so the transition was pretty straight forward.

With Google it was not a pure MySQL implementation as the company had engineered the system to suit its own needs. What was required therefore, was the need to test all the patches and modifications they had made, using MariaDB.

My take

This is a good example of why the commoditisation of technology will move the decisions about what users `want to do’ to a higher level of abstraction away from the mechanics of doing it. Selling the best `wrenches’ now has a limited life span.

Disclosure - at time of writing, Oracle and SAP are premier parents of diginomica.