2 min

Tags in this article

, , ,

Google has unveiled a new API, the Cloud Inference API. It is a managed analytics service capable of executing hundreds of thousands of queries per second.

The service is designed to process time series data, reports Silicon Angle. This is information that consists of several individual measurements that are sorted in chronological order. This way of collecting data is useful for sales analytics, where it can be useful to have insight into changes that occur over a longer period of time.

Google created a separate query language especially for Cloud Inference API. This allows users to indicate which time periods they want to focus on in an analysis, but also to correlate different datasets from the same period. The service allows users to investigate long-term trends as well as isolated incidents such as a day when more people came into the store.

Trillion data points

The service claims to be able to run queries against trillions of data points. Some queries can also be executed in real time. According to Google engineer Emanuel Taropa, this is something that is historically difficult for companies to implement themselves.

“Whether companies measure clicks, queries or sensor readings, they often generate time series or event-driven data. Analysing this data gives companies the opportunity to discover insights in real time. But often this also means that they have to create a learning system that can scale with millions or even billions of data streams. For many companies it is very challenging to design this.”


There are several applications for the real-time options of Cloud Inference API. For example, factories can analyze measurements of machine sensors when they have just been done. This allows professionals to identify potential problems quickly.

Cloud Inference API already has a number of customers, including Snap, the parent company of the social medium Snapchat. Snap concluded a 2 billion dollar cloud deal with Google in 2017. According to Peter Ciccolo, an engineer at the company, the service promises “to replace several specially made data flows with a single system”.

This news article was automatically translated from Dutch to give Techzine.eu a head start. All news articles after September 1, 2019 are written in native English and NOT translated. All our background stories are written in native English as well. For more information read our launch article.