You are currently on IBM Systems Media’s archival website. Click here to view our new website.

POWER > Business Strategy > Power9

POWER9 Servers Possess Automation Capabilities

Machine Learning

We’ve accumulated more than three decades of IBM i business data and system data, but are we any better off today than we were a decade ago?

Right off the bat, all of this data introduces a couple problems. First, the resources to really crunch the data in different perspectives are lacking. Second, botched data warehousing projects and failed business intelligence (BI) initiatives are running rampant in IBM i shops globally. Often, data is inefficiently exported to other platforms, which are stressed by the high volume of data.

Enter machine learning, an offshoot of artificial intelligence (AI). When coupled with the new IBM POWER9* GPU servers, we now have an affordable server that can incorporate machine learning into everyday business. The IBM Power Systems* AC922 started shipping in December 2017 and offers the price point and performance capabilities to handle these types of projects.

Now, you might be scratching your head as to how, when and why you would ever need this much processing horsepower or a GPU-based server. In a word: Google. Google has been gobbling up POWER9 processors for custom-designed servers to handle their immense search workload. It’s only a matter of time before the high-end technology that ships today becomes commonplace in mainstream computing—kind of an “if you build it, they will come” mentality. I expect more and more IBM i shops will find ways to use this technology very soon.

What makes machine learning different from past attempts at automation in the BI world is its ability to crunch large quantities of data in seconds using algorithms built by your data scientists to predict and project business outcomes based on data and today’s current situation. But how many of us have a data scientist? You just might have to grow your own to adopt this technology.

Security Applications

We even see the need in automation and security projects for companies. It’s becoming increasingly difficult to keep up with all of the information coming from just running the data center. The challenge there is configuring the software to have the proper business sense that meets your needs.

With a proper understanding and years of collective knowledge, a good administrator can configure security or automation software to have the thresholds they think are needed to react to events happening on the server, but that’s not good enough anymore.

In security, we need AI. Today’s organizations can’t continue to grow their spending and staff, and many struggle to get proper security programs in place. Image this: Your security software could look at your data, compare it to the regulations of the world, understand your industry, interpret the trends in security, and then configure itself to your specific business. Along the way, it learns what a bad day looks like compared to a normal day based on events coming from security probes across your servers and networks, in addition to known threats from the outside world. This scenario isn’t far off from reality.

Improving Business Processes

With machine learning, the industry is embarking in the next generation of solutions that can learn your environment’s data and apply the techniques of machine learning to learn from voice commands, learning from external information from sources like Google to help automate and secure your systems. Regulations and best practice documents on the internet could be used to apply rules that meet your industry’s requirements. We should see more of this as we move along in 2019.

Now, consider your automation software. Today, you load the software and must apply rules for monitoring different events, backing up your system and automating the schedule. With machine learning, the idea would be to load the software and let it determine the thresholds and rules that should be built to properly manage your servers.

Your automation software would also compare your system and application availability against industry benchmarks, giving you dashboards to better manage your day-to-day operations. IT operations would finally have real visibility into the impact system availability has on the business.

Another potential application: Machine learning technology that can identify bad SQL queries from an end user as something that always causes problems, resulting in a halted process or an alert to operations, and interrupt the statement before it can run.

The question of whether your server is performing well is always tough because no two IBM Power Systems* environments are the same. Machine learning applied to operations data could result in dynamic thresholds that are built around the history and requirements of your unique environment. Simply put, if you have one IBM i server running an ERP from Infor and another running trucking software from TMW, it won’t matter. Machine learning can adjust its knowledge based on each client’s environment.

Preparing for the Future

So, we imagine a day where machine learning would assist in configuring our tools and making tweaks as it learns on the job. But what can we do today to make machine learning more impactful tomorrow? The solution is simple: Talk with your vendor and implement automation, security and business intelligence solutions as soon as possible. You’ll start seeing the value add up immediately after implementing these solutions, and machine learning (when it’s ready) will easily layer over the top of these tools and use your past to build better rules for the future.

Tom Huntington is vice president of Technical Services at Help/Systems Inc. Tom can be reached at



2019 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.


POWER9 Servers Possess Automation Capabilities


POWER9 Brings Changes to the HMC


POWER9 Hardware Supports On-premises Cloud Deployments

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store