Google Unveils Many Advances in Open Data and AI at Cloud Next

In the context: When it comes to technical products, concepts are often more elegant than reality. Features and functions that seem logical and simple often turn out to be much more complex or time-consuming than they seem at first glance.

Part of the problem, of course, is that many of the most advanced technologies are complex and can be very difficult to implement. But an even more common problem is that pre-existing requirements are not fully explained, or the number of steps required can be much more complex than it first appears.

Simply put, the devil is in the details.

This applies to many cloud and artificial intelligence technologies. High-level product ideas, such as the ability to quickly analyze any type of data to help build artificial intelligence (AI) or machine learning (ML) models using new types of hardware accelerators, have been discussed for years.

As made clear by Google through several ads on its Cloud Next events, however, there are many important details that must be in place for these ideas to become a reality.

To begin with, not all data analysis tools and data platforms can work with every type of data. This is why the ability to import or ingest new and different format data types into a wider range of analytics tools is so important. Opening up the ability for data platforms like Elastic to access data stored in Google Cloud and Google providing support for Elastic in its recently expanded line of Looker business intelligence tools are just two of the many announcements related to open data. made in the Cloud. Next.

Similarly, different types of data are often stored in different formats, and analytics tools should specifically include support for these data structures to make them more useful to a wider range of users and application developers.

For example, in the growing field of data warehousing, where large “lakes” of unstructured data such as video and audio can be queried using the tools found in structured data stores, the open source Apache Iceberg table format is becoming increasingly popular.

That’s why Google has added support for this and other formats including Delta and Hudi to its BigLake storage engine and added support for unstructured data analysis to its BigQuery data analysis tools. This not only provides additional flexibility, but also means that unstructured data can use other Google Cloud Platform (GCP) Big Query tools, including machine learning features such as speech recognition, computer vision, word processing, and more.

Another important area of ​​development is related to the use of various types of hardware accelerators to improve the performance of the AI ​​model. For example, Google has created several generations of TPUs (Tensor Processing Units) that offer important benefits for applications such as AI model training or inference. In addition, there have been many recent announcements from well-known semiconductor companies such as Intel, AMD, Nvidia and Qualcomm, as well as many chip start-ups targeting this growing area.

As you might expect, each of these chip companies uses different methods to speed up AI and ML models. What is not so clear is that the methods required to write software or create models for various accelerators are also proprietary. As a result, it can be difficult for software developers and AI/ML model builders to take advantage of these different chips due to how difficult it can be to learn all these unique approaches.

To address this issue, a couple of more intriguing Google announcements from Cloud Next are the launch of a new industry consortium called the OpenXLA Project and the debut of new open source software tools designed to simplify the process of working with several different types of hardware accelerators. .

OpenXLA is designed to increase the flexibility of choice that AI/ML developers have by providing a bridge between the most popular front-end frameworks used for building AI models, including TensorFlow, PyTorch and JAX, and a variety of different hardware accelerators. Initial releases of software tools include an updated XLA compiler and a portable set of machine learning computational operations called StableHLO.

Companies that have also joined Google’s initiative include Intel, Amazon Web Services, AMD, Nvidia, Arm, Meta, and others. Intel’s inclusion is interesting because in many ways the goal of the OpenXLA project is similar to Intel’s own OneAPI, which aims to allow developers to use multiple types of Intel computing architectures such as Habana Gaudi CPUs, GPUs, and AI accelerators. without having to learn how to program for each of the different types of chips. OpenXLA takes this concept to an industry-wide level and, by including many key players in cloud computing, should open up a number of important new opportunities and accelerate the adoption of hardware accelerators.

As with many other announcements made by Google on Cloud Next, the real benefits of the OpenXLA project and related tools will take some time to materialize. In the big picture of tech industry trends, these tools themselves may seem a little modest. However, taken together, they represent very important steps forward and are indicative of Google’s efforts to make its tools more useful to a wider audience of people.

They also reflect a strong focus on open source tools and a desire to make their Google Cloud Platform and related offerings more transparent and flexible. The process of using all the technology tools that Google has to offer is still undeniably difficult, but with the extensive collection of announcements the company has featured on Cloud Next, it’s clear that the company’s evolution as a major cloud provider continues to evolve.

Bob O’Donnell – Founder and Principal Analyst Technalize Research LLC technology consulting firm that provides strategic consulting and market research services to the technology industry and the financial professional community. You can follow him on Twitter @bobodtech.