2020 The Year of AI Open-Source Software in the Mainframe

As we are approaching the end of an unusual year, a difficult 2020 with a global pandemic and high unemployment that has disrupted the lives of so many people, I’m reflecting on some of the positives that we can take from this year. In my world of technology and open-source software, innovation didn’t stop; in fact, we can argue that there was an increase in productivity by having millions of people working from home, reducing commute times, travel, and unnecessary meetings.

Software innovations are happening in the open; yes, this year, again, most of the latest innovations are open-source software projects built with one or many other open-source software components. Augmented reality, virtual reality, autonomous cars, artificial intelligence (AI), machine learning (ML), deep learning (DL), and more are all growing as open-source software. Needless to say, all programming languages and frameworks are open-source, too. Open-source building blocks such as PythonTensorflow, and Pytorch to name a few, are powering the latest innovations.

I like to keep an eye on the growth of the different open registries and repositories. GitHub has surpassed 100 million repositories and more than 50 million users this year. NPM, where JavaScript/Node.js open-source packages are available, surpassed 1.4 million packages; Nuget for open-source .NET code surpassed 220,000 packages; and Python packages available in PyPI surpassed 270,000 [1]

The number of open-source projects in the AI and Data space is growing exponentially. It is now hard to create categories to classify all the open-source software available in this space, take a look at the LF AI & Data foundation landscape or FirstMark AI and Data landscape for a sample of software available in this space.

With a growing number of open-source software to create AI applications, we also have an increase in real-life use cases. Businesses across industries are adopting AI to address real business challenges and opportunities. Healthcare providers using ML and DL for faster and better diagnoses, telcos using AI to optimize network performance, the financial services industry reducing fraud, and generating better predictions are just a few examples of use cases we see now every day across every industry vertical. 

There are many more examples to add for insurance, transportation, government, and the utility industries. One common denominator across these important industries is that all have mission-critical applications with very valuable data running on mission-critical platforms.

Mainframe platforms host the most crucial business functions in all of these industries. For decades, they have continued to improve their technology in high-speed transaction processing, capacity for very large volumes of transactions, best-in-class security, and second-to-none resiliency.

When enterprises need AI applications in the best platform for I/O intensive transactions of structured or unstructured data, there is an ideal mission-critical platform; when AI applications need high-performant access to storage and databases, there is an ideal mission-critical platform; when AI applications need to secure data in transit, at rest and in use with confidential computing, there is an ideal mission-critical platform; when AI applications need a resilient platform that provides 99.99999% availability, there is an ideal mission-critical platform designed to deliver on all of these criteria.

Mainframes are this ideal mission-critical platform that can tightly integrate AI, ML, and DL applications with data and core business systems that reside in the same platform. In other words, they provide a secure high-performance environment to bring AI, ML, and DL to existing transactional applications and deliver real-time insights and predictions.

The ecosystem of open-source software for mainframes (s390x processor architecture) continues to grow. I believe it is at its best in 2020, and I have great hopes for the upcoming 2021 to be a year of continuous growth in the open-source software ecosystem for this mission-critical platform.

The most popular open-source software for AI has only existed for a few years. As we are coming to the end of this difficult 2020, we see that it has been a strengthening year for many open-source projects. Tensorflow and PyTorch are used more than ever, and a number of open-source projects are becoming very popular, for example, Egeria, Pandas, Jupyter Notebook, Elyra, ONNX, Kubeflow, and others that I hope will continue to grow and be available across all platforms in 2021.

Open-source is not a trend; it is here stronger than ever. We are going to continue to see innovation and enhancements in the AI and Data open-source ecosystem. The data that resides in mission-critical platforms such as IBM Z and LinuxONE is a valuable asset for businesses and can be used for creative AI solutions.

AI open-source software and mission-critical platforms introduce exciting possibilities in 2021 and beyond.

[2] Free image by iXimus from Pixabay


[1] Source: Nov 2, 2020 www.modulecounts.com

Editors Note

This blog post was written by Javier Perez, Open Source Program Manager, IBM I suggest you check him out on LinkedIn

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.