Whenever I’m thinking about a new company or tech breakthrough, I ask myself how it helps us get to a world that can support a few billion or even a trillion devices. One obvious challenge is the current cloud computing model, where information and data is centralized and then pushed to devices.
As I talked about a few weeks ago, there isn’t enough room in current or future cloud computing infrastructure to support all of the data generated by connected products. The answer is edge computing, but for true computation at the edge, we need someplace to store data.
To solve this problem, Berlin-based startup ObjectBox has created a distributed, low-resource database that can synchronize information from different devices. There are several different elements here, so let’s tackle them in order of coolness.
ObjectBox can run on devices that don’t have a lot of memory or computing power. It uses a client-server model of information exchange, with the client devices requiring 1 megabyte of memory to run the database. Smaller devices, such as sensors, only need about 50 kilobytes of memory, says CEO Vivien Dollinger.
ObjectBox’s database is a NoSQL database that will compete with SQLite, an embedded database popular with IoT projects. The ObjectBox database memory requirements are incredibly low, which means you can put the database in everything ranging from an Intel-based edge gateway to a temperature sensor running an ARM microcontroller. When it comes to how it handles synchronization, it’s easier to think of the database as almost a client-server model.
The ObjectBox sync feature only offers synchronization of pre-selected data, as opposed to a true replication of all the data stored in the database. But it’s enough to handle things like the state of a device without having to request that information from the cloud. Dollinger says a developer would have to select which data gets synced across all of the devices, and that it’s more of a “pull” model than a “push” model.
That means the smarter versions of the database will pull the information needed from sensors when needed. For Dollinger, this tech solves a practical need for the internet of things and also addresses one of her largest concerns about it: privacy. Because a device’s state can exist outside of the cloud, it’s possible to envision a group of home or enterprises devices that communicate locally on their own intranet.
So, you could tell Alexa to turn on a light and the Echo would pull the current state of that light and command it to turn on — as long as it was off without every contacting the cloud. Google is working on such local control for its smart home products, but I’m not sure how its Local SDK for Google Home handles device state.
Dollinger says that while the database is open source, ObjectBox plans to make money by implementing features (such as sync) on top of it. The company will also charge for training and support. Let’s see where it goes.
Updated: This story was updated on Tuesday, September 17, to correct the fact that Object Box is not based on SQLite.