IBM InterConnect 2017 closed its doors last Thursday. It felt like school letting out for the summer with fewer people walking around the concourse with the feeling of “we are done!”
This was my first InterConnect conference and I found it very interesting. It was the typical big-tent IBM conference that needs the largest venues in Las Vegas to accommodate it. So, what I did I learn in my four days? A great deal—some good, some a little disquieting.
Watson, Bluemix, Blockchain
IBM’s CEO Ginni Rometty and her company were pushing three things: Watson, Bluemix and Blockchain. She has pushed all of IBM’s chips into the center of the table at the expense of other parts of the organization that have been historically successful for IBM, such as the mainframe. I won’t prognosticate as to the accuracy of her bets.
The one thing you could not escape at the conference is Watson. Yes, Watson is everywhere and, what I learned is that Watson requires data and even more data to deliver the level of cognitive computing that IBM is promising. This sounds great if you are an optimist. After all, IBM does claim to be building a “smarter planet.”
If you saw the Terminator and know about Skynet or if you were a viewer of the recently cancelled CBS TV show “Person of Interest,” this probably scares the heck out of you. I’m not sure where I land on this, but it’s pretty easy to see how someone could use Watson for nefarious purposes. This bears watching over the next several years. I have little knowledge of blockchain, but those with more are skeptical of Rommety’s chances on this. She sees it as the Internet redux—which Al Gore invented, right?
Neither Watson nor blockchain were my primary purpose for attending InterConnect. It was to get some insight on the mainframe and DevOps.
The Problem with IBM’s Mainframe-DevOps Approach
Several IBM (and Compuware) customers presented on their efforts along with IBM. When the companies talked about the tools used in their DevOps toolchain they were mostly colored blue (i.e. IBM tools). When IBM talked about DevOps they painted a slide of completely IBM tools. There was some lip service paid to Jenkins, but when questioned the IBM Offering Manager slammed it pretty hard.
Git was the other tool not made by IBM that was mentioned most often. Git is an extremely popular open source tool for source code version control. Rocket Software ported it recently to z/OS. Anyone can download and use it on the mainframe, but Git’s future on the mainframe remains to be seen.
The problem with IBM’s true blue approach to the mainframe DevOps toolchain is that you might end up with competing products within a company and upset a majority of your developers by forcing them to use a different set of tools than they are used to.
The other important information I took away from the conference was around APIs. Everyone was talking about RESTful APIs.
Jason Fournier from Verizon, a twenty-something developer, gave a fantastic presentation how he and his team had developed REST APIs for their partners to call. Prior to the REST APIs being available the partner would make a request for data and someone would have to extract it from the mainframe. Now they publish the API, and along with some authentication the user can self-service the request themselves. They charge for this data. Every time an API is called or a specific amount of data is downloaded, a bill is sent.
Jason gave a quote that I’ll remember for a long time. When talking about how much they were doing on the mainframe, he said, “If we slow down, we will fall behind.”
Another session on APIs led to a pretty lively discussion. Some more senior people talked about all the interfaces as MQ or CICS SOAP calls they could make, and another twenty something said “If you force everyone to use antiquated interfaces to the mainframe, they will look elsewhere for the data.”
These discussions on APIs got me thinking about more APIs for Compuware products I manage and gave me the ability to convince skeptical developers why we need to create them.
The final piece I’ll talk about is IBM’s version of a cloud for the mainframe called zCloud. There were at least two sessions on the zCloud. After listening to Steve Dickens talk about it and describe it, I didn’t see how they could ever deliver the savings promised, and it was not much different than outsourcing.
I don’t really understand the point of that either. Most companies today are more than ever dependent on the technology that runs in the datacenter, and when you add a third party that prevents you from going faster, you need to question the need for it.
All in all, it was a successful week. It was not all hard work. IBM threw a party on Wednesday night with Andy Grammer and Zac Brown Band. I had limited knowledge of both but came away a fan. Now I’m catching up on everything that happened last week while I was buried in DevOps, Watson and APIs.