IBM is bringing machine learning to its flagship mainframes.
The company said Wednesday that IBM Machine Learning, a platform for training models and ingesting data, will be available on its z Systems portfolio via its z/OS.
IBM's take is that mainframes hold much of the enterprise data globally so bringing machine learning to transactions makes sense. Financial services, retailers, insurers, and governments use mainframes heavily. An IBM z Systems mainframe can process up to 2.5 billion transactions in a day.
According to IBM, its machine-learning platform will first be available on z/OS and its mainframes, but it will also be on IBM's Power Systems and other platforms.
IBM Machine Learning supports any programming language as well as multiple frameworks such as TensorFlow and Apache SparkML. The platform is also using data science automation tools created by IBM Research.
Big Blue plans to hold a launch event Wednesday with Rob Thomas, general manager for IBM Analytics, and a bevy of other executives. Customers will include Travelers and America First Credit Union.
- IBM launches cognitive computing hardware unit: Enter the Watson, Power 9 stack
- IBM adds TensorFlow support to its PowerAI
VIDEO: This is IBM's five-year plan to remake healthcare