Machine Learning (ML) is a powerful predictive tool for all enterprises. ML has become indispensable for the business side and enterprise IT systems by detecting patterns and taking advantage of highly likely outcomes. In a data center or cloud environment, ML could predict the performance of an application server or machine well before a conventional monitoring system, give advice on future cloud billing based on prior usage, recognize patterns in application data access to optimize performance, etc. There is a growing number of use cases for applying ML to IT systems and DevOps in general. Because of this need, standard interfaces between IT systems and ML engines will only grow in importance for all enterprises.
Zoran Sevarac and Frank Greco, both Java Champions, are the co-authors of JSR #381, Visual Recognition for Java. This standard Java API offers an API using Java-friendly conventions that is comfortable and familiar to all Java application developers. It also avoids many of the issues of other Java ML APIs that either thinly disguises a C/C++ API or require deep data-science expertise. There are several implementations of JSR #381, including Deep Netts and Amazon’s DJL (both OSS).
Eclipse Gran Sasso predicts performance of cloud-native enterprise Java applications and traditional application servers using AI/ML techniques. By building deep learning models and using associated ML tools, we will be able to prescribe optimal resource allocation and costs for given user loads.
Eclipse Gran Sasso is a pilot project that predicts performance of cloud-native enterprise Java applications and traditional application servers using AI/ML techniques. By building deep learning models and using associated ML tools, we will be able to prescribe optimal resource allocation and costs for given user loads.
This pilot will be an open-source project under the governance and sponsorship of the Eclipse Foundation. We are looking for partners to provide performance data, benchmarks, and specific use cases as part of this pilot. Our long-term goal is to establish standard, open interfaces between ML engines and IT infrastructure.