Is open source technology a disruptive innovation force in the Clayton Christensen sense of the term — meaning it starts out serving the underserved or unserved, and works its way up to eventually chase the market leaders upstream?
It certainly meets that definition, in that it has made powerful solutions available to the smallest and budget-tight players, and, yes, many open source solutions have pushed their way up the food chain to eventually challenge well-funded commercial competitors. We see Linux in many Fortune 500s, and Apache Hadoop is no stranger in these places, either.
The database market is another example of open source solutions pushing their way from the bottom to the top. I recently had the opportunity to chat with two of the leading protagonists in this space, Monty Widenius, creator and co-founder of MySQL (now part of Oracle) and currently CTO of MariaDB, and Michael Howard, CEO of MariaDB, to talk in detail about open source and cloud. MariaDB is an open-source database company that picks up where MySQL left off.I was invited to their first user conference in New York,
The following is a mashup of my Q&A with Widenius and Howard:
Q: What’s your relationship with MySQL?
Widenius: The project started in 1981. I created a database that was useful for doing applications. MySQL was started in 1996. Maria is more MySQL than MySQL. We were acquired by Sun, then in middle of the process, Oracle announced they will buy Sun. So most of the technical people left and joined me.We are the same people who created MySQL. We are the new M in LAMP. .
With MariaDB, I’m still coding, and wrote many of the new features. I always believed it was easier to find good managers than good coders. I’m a lucky man — I spend 60% of my time coding.
Q: Do you see open source solutions — such as MariaDB — as a disruptive force?
Howard: Yes and no. There is some open source componentry, which is just more of the same. The notion of cannibalization or disruption is more nuanced. you take something like Oracle which owns a bunch of databases, and they will prioritize based on margins or profit, which is what a company should do. So lower-end products wont get the same features as higher-end products, such as data warehousing or analytical features. So that puts them in a difficult situation if someone comes in with a very cheap data warehousing vehicle.
I think we have taken it one step further in the case of disruption. In the case of analytics, we’re making it a feature in our database. So in the entire $18 billion market, we just featurized analytics. Even though Amazon is incredibly efficient, incredibly automated, they can’t keep up with the innovation they’re in.
Q: What’s your view of the cloud model? Do you see it as an effective delivery mechanism?
Widenius: We are in all the cloud. Microsoft Azure, Alibaba, and lots of big vendors are making a lot of money. But after you start a company, some times its eventually cheaper to run things in house. I don’t believe cloud is the solution. There lots of vendors who don’t trust the cloud for security reasons. Its hard to not be in control. The cloud vendors are really not open source friendly either — they’re users, and sometimes misusers. Private clouds are the future. I believe for mission-critical things that have to be secure, you need private cloud.
Howard: Some of these cloud vendors — such as Amazon and Alibaba — are among the most valuable companies in the world. They’re seeing incredible profits, incredible hegemony, and incredible reach. You can’t just put them in a box in the corner, and not see what they’re doing. They make you work a little bit harder, and you have to really be ahead of them.
I do believe, however, there is tremendous growth and tremendous opportunity in on-prem, and that will even be fortified as time goes on. As.storage technology network and compute vendors take advantage of technologies that’s faster and at price points that will make Amazon look expensive. Those technologies are here today, they’re just not, in my book, packaged appropriately. Cloud is here to stay; it’s a wonderful check and balance for on-prem vendors in ways of thinking about deploying applications.
Q: Do you see more functionality going to small-footprint mobile devices?
Howard: We have some people putting our database on a tablet. But the way you normally do something like that is to have your Salesforce icon, or some kind of, the veneer of your application is on your iPhone. Then you run your bigger loads on your server. For the end user application, why not? But if you’re developing, that may not be the best form factor to do that.
Q: We’re rushing headlong into machine learning. Can we trust the data?
Howard: You need to be concerned how solid the data is. Yet you hear about the automation bias of the machine learning. Theyre all black boxes. You can’t go into the system of the algorithms and understand how did it actually come up with that answer. They don’t know the why. Everyone is talking about AI, yet they cant understand how the statistical models behave. There’s no mandate on the data going into the models. They’re back testing, announcing 70 to 80% reliability of whatever, who knows what that is? We need a Moody’s of data, but not 2008 style. That’s why the database has to take that role — because it’s extremely objective, and it doesn’t have a back box.