IBM i And Watson: Another Step In The Right Direction
March 13, 2017 Dan Burger
One way to understand the future of enterprise computing is to focus on data and systems integration. Not everyone likes that idea. There are strong feelings about certain platforms and processes. But as the IT-business alignment evolves, integration will prevail. The current emphasis, driven by data analysis and cognitive technology, is a powerful motivation.
IBM’s Watson is on the big stage. Explaining Watson’s analytical power and how it can be integrated into business computing is one of the main cognitive attractions. It will take serious planning and incremental steps to bring Watson into widespread use. One of those incremental steps has been worked out by two technicians at Sirius Computer Solutions.
Their proof of concept involves an IBM i integration using a messaging system known as IBM Message Hub. The two technicians have a ton of IBM i integration experience. Jeff Whicker has been at this for more than 26 years. MQ and Message Broker are in his sweet spot. Matt Shannon started as a Java developer more than 15 years ago, and has deep experience with WebSphere. Many of the customers he works with are IBM i shops.
The proof of concept wasn’t created for a specific business, but it is a real concept that demonstrates how an IBM i integration with Message Hub on the IBM Bluemix cloud, can be useful in multi-platform operations. More importantly, it’s a first step to reaching Watson integration with IBM i. Depending on the infrastructure on the IBM i system, relies on methods such as SQL calls to DB2 for i, program calls, and data queue calls.
“We’ve always been able to do the same stuff with MQ or other methodologies,” Whicker says, “but this shows an option to on-premise messaging.”
IBM MQ is messaging middleware used to integrate applications and business data across multiple platforms, regardless of whether they are on-premise of in the cloud. It’s being used to obtain real-time insight from mobile, Internet of Things (IoT) and sensor data.
Message Hub is IBM’s cloud-based, platform as a service technology designed to stream batch and real-time data to analytics applications.
IBM i-based companies beginning the integration of diverse applications and business data across multiple platforms will want to consider Message Hub, particularly if advanced analytical requirements are a priority.
Whicker says one of the points in the POC was the capabilities to do analytics on the cloud. “Once you have a message going through IBM’s cloud, you now have Watson to do analytics if that’s what you want to do,” he says. “Once an organization sees that messaging is working well in the cloud, it becomes easier to expand. Rather than writing an exit program in MQ, there’s an option to talk with the database – create files so as messages are processed, they are summarized in these database files.
“Existing companies are often busy maintaining code, but they are curious about the cloud and don’t know how to get there,” Shannon says. “I felt the same situation myself. I know there’s a lot going on that I am not a part of. Doing this simple concept has led me to see the possibilities. You can take a message that you’ve been sending, but now connect to Watson for analyzing components.”
There will be learning curves. New technologies, like puppies, require training. Stop learning and you find yourself in the dog house.
For some companies, it begins with moving to modern RPG, which includes modular development techniques, ILE, service programs, and Web services. Incremental transitions are less disruptive than huge leads. Just like when Java came along, it was an easier transition for those who had modern RPG skills. Now it’s Node.js and Python that are transition destinations. IBM i has a pretty good tool kit for Node.js developers, with hooks into the OS and the DB2 for i database. And Node.js developers already have their fingerprints all over the cloud.
A lot of the cloud connections are using Node.js, Shannon notes. “Node has a lot of momentum.” The open source community is building modules for a variety of purposed, so developers can pick up and use code that they don’t need to write themselves, which is convenient and speeds the development process.
“Companies with modern skills, strong core code, and a modern Web interface are in great shape for moving into the cloud for integration,” Whicker says.
The IBM Bluemix cloud already includes an impressive array of available APIs. Some for Message Hub, IoT, and Watson. Some APIs will need to be learned, depending on the direction an organization chooses, and there are many potential directions when you consider all the major platform choices that could figure into the integration.
“We’ve seen all types of shops take on integration projects in the past,” Whicker says, “including a two-man shop that hasn’t had time to get off V5R4.
“The shops’ current status and skill sets in place are not as important as whether the RPG is ILE and modular. We’re not talking about new processes. We’re talking about new implementations. And a more simplified way of talking with vendors and clients. And a more simplified way of adding platforms and infrastructure. Anything done to be more modern is going to help prepare for this.”
Planning an integration project like one that integrates IBM i and the Bluemix cloud will require a great deal of flexibility and uncertainty. That makes it pretty much the same as any other project whether it’s IT related or some other workflow process.
One part of the Sirius proof of concept project involves getting data from an Internet of things (IoT) center.
“There’s a lot of talk about the IoT and the amount of data it will generate,” Shannon says. “What kind of devices will handle that data is something to consider. We move it to the cloud to spread the workload. There needs to be a strategy for processing data. The IBM i is very powerful and can do a lot of processing. But companies will pick and choose where the processing takes place.”
The infrastructure to handle an integration project like this one seems to lead to the cloud having an advantage over an on-premise option. IBM’s infrastructure as a service and platform as a service means infrastructure is easier to add and costs aren’t a big up-front hit for companies. Eventually we will likely see an IBM-sponsored total cost of ownership (TCO) analysis that is favorable to IssS and PaaS.
“Looking at it from an i developer’s perspective, we wanted to tie in to our existing legacy systems as unobtrusively as possible. We didn’t want to have to rewrite programs or cut out the guts of programs,” Whicker says.
And Shannon adds, “We realized businesses go on every day and we didn’t want to change things that don’t need to change. But we do want to be able to do new things.”