Utimaco, the well known cybersecurity firm based in Germany, recently launched its much anticipated Hardware Security Module (HSM) monitoring & management software, u.trust360. In a world where businesses are accumulating increasing amounts of data, cyber security is becoming more and more prevalent, and Utimaco eliminates this problem through its wide-array of HSM products and management softwares. Essentially, the HSM is a physical device capable of encryption/decryption and key management, widely used in the Payment Card Industry (PCI), handling anything from payment information to sensitive company or employee data. Previously, these HSM systems could experience reduced speeds or freezes during peak seasonal times for businesses where the system is functioning near or at full capacity. What makes u.trust360 unique is its integration of machine learning models to predict seasonality and peak usage of the HSMs for precise monitoring of existing devices and accurate forecasting of seasonal peaks.
Leveraging open-source data, four Berkeley students conducted research to effectively create a machine learning model to accurately forecast future peaks in CPU usage of Utimaco’s HSM devices monitored by u.trust360. This integration will provide Utimaco customers with profound insights into future device usage, alerting customers when storage capacity is expected to reach its maximum, and allowing them the necessary time to respond accordingly.
New product launches can be tricky and in order to be successful and stay ahead of competitors the product must be cutting edge. Utimaco does just this with the u.trust360 through implementation of all new predictive capabilities within the user-interface of the device. Using both past and current HSM data, this new machine learning algorithm will provide customers with analytical insights of trends and seasonality and based on the trends send alerts days ahead of potential peak times, to avoid reduced server speeds and possible system shutdowns. These capabilities will allow customers to effectively utilize Utimaco’s new management capabilities of u.trust360 to operate HSM devices to the fullest extent without the headaches of dealing with spikes in usage and the possible negative effects to their customers that ensue. This is a first in the industry, and Utimaco ultimately hopes to continue being an industry leader in incorporating machine learning and predictive analytics into all products.
Berkeley students were able to achieve this feature through researching and utilizing time series modeling to predict usage data, trends, patterns, and seasonality. The team is confident their time series models will be scalable to any Utimaco encryption device in the future.