The Server Transformers

Founded in 2012 by the supercomputing experts Emilio Billi and Antonella Rubicco, A3cube pioneers the transformation from the high-performance computing systems to high performance data systems. The company strongly believes in the importance of providing better equipment to the community to extol the maximum value from all the data collected, this makes A3Cube focused in the development of different complementary technologies that all together can build innovative and market-unique solutions.

The holistic approach that A3Cube takes starting from the analysis of the bottlenecks in the existing hardware in the relationship with the applications architecture was the basis for A3Cube system orchestration. Through their technology, the company can build a system that perfectly fits the application needs maximizing the application performance without needs for special application tuning. The result is that an ordinary server in combination with A3Cube composable infrastructure can achieve the maximum potential outperforming any other equivalent system in the market.

" We combined all the cutting edged features in a single robust product that can fit the needs of any applications "

According to Antonella, founder, president and CEO of A3Cube computing is only two things “data” plus “math” over the data, this simple concept is at the basis of any computation from simple word processing to gaming to the most intense HPC challenges to AI and so on. In HPC system architects are mainly focused on providing fast “math” resulting in faster computing. In the modern data-driven era the “math” is performed over a vast amount of data, so the key is to provide not only fast “math” but also fast “data” access to have the right amount of fuel the “data” to perform the computation at the maximum speed possible. Imagine that a standard modern CPU can perform 800 operations every 100 nanoseconds in other words this means that every 100 nanoseconds that the system wait for a data firms lose 800 transactions. With A3Cube architecture the time for the data is reduced to microseconds making the data access faster and the computation more intensive by increasing the number of operations that the system can run 100 times greater than in standard systems.

A3Cube observed that all the existing storage systems are or were based on traditional NAS or were backed by a parallel or distributed file system specifically designed for an application. And, if one needed to be flexible and combine servers in many different ways, no robust solution could be integrated without pain. Starting from these observations the company decided to invent a brand new way to orchestrate the data access in multiple scenarios using multiple file systems combined and embedded hardware tricks that permits everybody without any specific experience to configure and build robust systems extremely optimized in minutes. “We combined all the cutting edged features in a single robust product that can fit the needs of any applications,” adds Antonella. “We create an infrastructure that is deeply integrated with the network infrastructure, the IO bus of the server and the related attached devices.”

In an instance, one of A3Cube’s customers had the challenge to predict the status of a service in less than 60 seconds analyzing more than one millions of records from multiple sensors per time and performing some mathematical operations on them to decide if the data were valid or not. In the case that the time to analyze the data was higher than 60 seconds, the system considered the data wrong as default, creating a security issue and downtime in the chain. A3ube’s real implementation exceeded the expectation by analyzing the same amount of data in less than 25 seconds permitting to implement a double check and a new series of controls over the data. A3Cube is growing at an unprecedented way and for the road ahead the company is developing new evolutionary products like an AI-driven scanner for materials quality analysis, and the new GRIFO™ for automotive algorithm training.