On this episode of the “I Am a Mainframer” podcast, host Steven Dickens is joined by Shuang Chen, PhD, a seasoned entrepreneur, senior technologist, inventor, and recognized industry leader. Shuang brings over three decades of industrial experience in technology innovation, software, large-scale internet systems, entrepreneurship, and business executive leadership.
Previously a senior research and development leader at IBM TJ Watson Research, the founding CEO of an internet venture, and the Chief Engineer of the world’s largest e-ticketing system project (designed for 5 billion tickets annually), Shuang’s experience is vast. He also launched the first and largest e-ticketing cloud service for intercity bus operators in Asia — a service that averages 20 billion tickets per year!
Shuang received the “Top 50 Chinese Americans in Business” award in New York State, holds a PhD in Computer Engineering from Rutgers University, and studied EMBA at the Wharton School at the University of Pennsylvania. Shuang is also the author of 20+ patents and a frequent speaker at global industrial conferences.
During their conversation, Steven and Shuang talk about his current role at Weather Corporation, how he got his start in the mainframe space, and his early work in AI with pattern recognition, image recognition, voice recognition, machine vision, and core machine intelligence. Shuang also encourages everyone to learn and understand AI and deep learning because it’s much more than a trend.
Watch the episode on YouTube:
Listen to the episode on Spotify:
Transcript
Announcer: This is the I Am A Mainframer podcast, brought to you by the Linux Foundation’s Open Mainframe Project. Episodes explore the careers of mainframe professionals and offer insights into the industry and technology. Now, your host, senior analyst, and vice president of sales and business development at Futurum Research Steven Dickens.
Steven Dickens: Hello, welcome to the I Am A Mainframer podcast. I’m your host, Steven Dickens, and I’m joined here today by Shuang Chen from the Weather Corporation. Hey, Shuang, welcome to the show.
Shuang Chen: Thank you, Steven. Happy to be here.
Steven Dickens: Let’s dive straight in. Tell the listeners and viewers here a little bit about your current role and what you do.
Shuang Chen: Weather Corporation is a new venture and just a little over one year old, but this would be the third large project for me using the mainframe, the latest computing platform. Weather Corporation is working on a very exciting project. We will roll out sustainable local climate adjustment solutions anywhere in the world powered by the most powerful computers and AI platforms.
Steven Dickens: Fantastic. Let’s maybe start here and get a picture of you. The show is called I am a Mainframer; I really want to understand a little bit about your journey, so maybe take us back, Shuang. You’re finishing college; you’re starting to enter the workplace. What does that first role look like, and where do you start your journey in IT?
Shuang Chen: Oh, yeah, I started the journey in IT a long time ago. When I finished college, I went to graduate school. I was in the AI area. So today’s AI, that was like 40 years ago, AI was very theoretical parts. Mostly pattern recognition, image recognition, voice recognition, and machine vision are called “machine intelligence,” and that’s where I started my first master’s degree. I have more than one, so my first master’s degree is on the breakthrough speaker, independent speech recognition, which was the first time I got into the AI and no neuro network. That’s very simple ones and, because of the computing capabilities, that time was very limited. But that was where I first got started.
Steven Dickens: Okay, so maybe what was that first career move? Where did you go first, Shuang? Where was the first company you worked for?
Shuang Chen: For the first company, this was after I… Let’s see. When I first got into industry research, that’s during the latter stage of my Ph.D., I did some intern work in Bell Labs in New Jersey, which was on computer vision and which is also my Ph.D. through this, that was, that’s first.
Then right after that, after my Ph.D., my first real full-time work, it’s in Silicon Valley. So I worked there for five years. Also, work on the first generation of handwriting recognition for the mobile device; if you remember, that was, yeah. Apple had a so-called “Newton,” and Microsoft called it the “Pen Window,” the startup I’m working with in Silicon Valley has become a third competitive provider along with IBM. IBM has its own. Then this company provided its handwriting software and then eventually licensed by all the large PC manufacturers, including IBM. I was leading those efforts, and eventually, the company went to NASDAQ and et cetera.
Steven Dickens: Fantastic, fantastic. So when did you first start working with the mainframe? When did you first get onto the platform?
Shuang Chen: Yeah, in the past 20 years, now it’s 2023, I’ve worked through several large mainframe systems, some largest systems in the world. So my first large project with a mainframe was from 2003 until 2011 or 12, and I was the chief engineer of one of the largest ticketing systems in the world.
Steven Dickens: Oh, okay.
Shuang Chen: Yeah, we use the mainframe, I think, at that time, the Z10 and the TPF. So I work very extensively with just about all the largest ticketing systems in the world, from airlines, United, American, and Delta to the GDSs, Travelport, and Saber, and also was the largest ticketing system or transaction system in the world, including Marriott, and Chantalia, SNCF, and even Visa.
Steven Dickens: So we don’t get many people on the show with a TPF background.
Shuang Chen: Yeah, that’s not the first one. Yeah. The second one.
Steven Dickens: Maybe tell us a little about your time working with TPF. It’s some of the biggest mainframe deployments in the world, but not many customers in that space, and pretty… We don’t get many people on the show who’ve worked in the TPF space.
Shuang Chen: Okay. TPF, I think, stands for, if I’m wrong, correct me. I think it stands for the transaction processing facility. It’s a very traditional acronym. However, as we know, the mainframe was built in the sixties, and late sixties, last century. The mainframe was not built as a theoretical computer. It’s built as some foundational economy-needed computing platform. So, the mainframe designed from the beginning is one of the enabled airline reservation systems, which is a ticketing system in real-time and retail. One of the largest ones is a Cirrus, right? Then banking is like a CICS, so we all know CICS. So TPF is one of them built from the right beginning. What I mean by the right beginning is that, at that time, the mainframe was not like today. So, you get an operating system; then you got a middleware; then you got an application, you got a front end. No. The mainframe was one computer, a monster computer with all the blue-screen terminals, right? All the green screen terminals.
So all the transaction doesn’t matter anywhere in the world, or all the travel agencies in wherever, or all the airline counters and backend connect to the mainframe. So the mainframe is designed from the bottom up, from every bit and the byte, bottom up for that purpose. So the most criterion for the real-time ticketing system is that you have to handle very large scale simultaneous transactions, and that means how many transactions per second and a very reliable, very fast, and also maintain, you can say five nines or six nines, whatever the availability is. So in practice, I know that some largest ticketing systems, even today, they’ve been using TPF still on the backend, with nothing to replace it. Because if you replace it, you’re going to do the same thing and just with another name. So they have, for years, for many years, zero downtime. There’s not even scheduled downtime when you do maintenance; when you do database consult, they do whatever you do. They don’t need to turn down the machine or stop the application. That was the mainframe. Very. The top of that is very secure.
So security, reliability, and availability are on top of that large amount of transactions. I can give you an example since you’re asking TPF. You’ve heard the word the PNR, right? And on all the airplane reservations today, you have a record number. When you make a reservation in the Travelport or Saber or something, you’ve got a record number, which people also call a PNR. That stands for “passenger name record.” PNR. So every time you go to order, you get the query, they book, they can order. So you have a PNR. In order for the mainframe to handle this in the fastest possible way, the PNR number is the address on the disc for the record data physically located. So you don’t need other data to look up a table or index database. No. You just get that data and get to the address and the order.
Steven Dickens: So it’s completely different… And I mean, I know this about TPF, I used to have the pleasure of sitting with one of the distinguished engineers who had a background in TPF, and he told me about it. From a system design perspective, it’s just completely optimized for transactional speed. It’s not using any of those constructs that you would expect in sort of classical computing.
Shuang Chen: Right.
Steven Dickens: Where you’ve got a database, you’ve got metalware, you’ve got a sort of transactional front-end layer. It’s really, as you say, purely designed to drive transactional speed. So let’s move forward a little bit here. So you spend some time on TPF; what’s your next role after that? Where does your career take you next?
Shuang Chen: Yeah, so we just quickly wrap it up. That was significant because we needed to build the largest ticketing system in the world, and the people had never seen that large. It is designed for 5 billion tickets per year. Now each ticket is involved in many transactions. It’s not just one transaction. You’ve got a query; you’re going to book. A lot of people query but never book. You’ve got to book, but many people who book never confirm, right? So, you have to pick the day to pick the trip departure and arrival and time on that, and you have to make sure the tickets are available. So, when you book, you have to confirm it. So once you book, the seats are reserved for you shortly, then you have to confirm. So those are many transactions. And 5 billion tickets. That’s translated to a huge number of transactions in the current market.
So basically, at that time, it was about 500 million tickets per GDS. There are three large GDS; there are several small ones. So if you add all the three large GDS together and the times three, also not even 5 billion. So you can see how big the ticketing system is. So in order to do that, the industry put the best together, including IBM and TPF solutions. There was a huge X86 solution, which is best for the computer from HP, called Superdome. And then the best software from Oracle, Oracle Cluster Enterprise. Then, that’s another system. The three largest solution systems you can put together, and then we ran about a six-month proof of concept with the same dataset and criteria on the three different sites and teams. They have to really run through the computational results. Then eventually, compare, and compete in that sense; it’s a large project. So TPF wins out. That says something about this fundamental system built on the mainframe.
So the second one was based on that. Later, during these 20, let’s see, 2014 to 2018, 2014 to 2017. So that was the first time when LinuxONE was announced, and at that time, we were building the largest e-ticketing cloud service for the bus industry. The previous one was for the rail, and this is for the bus industry. The bus industry is even bigger. It was 20 billion tickets a year industry. But we’re not trying to build a system for 20 billion tickets. We break out for each provincial market roughly 1 billion tickets a year. Still quite large, but mostly is that you’re going to have, instead of one customer, you have many customers, hundreds of operators. So we provide a cloud service. They don’t have to do anything; they just open the interface for their ticket inventory every day. Because they change all the time. And we’re running in their brand about all the transaction and the sales channels and the backend, the transactions because you have to maintain the high integrity of this ticketing inventory because you cannot sell one seat twice or more than once.
So we run websites, mobile apps, like iPhone and Android, and ticketing machines like kiosks, and even we run these boarding machines, which are needed to check the tickets. So once the ticket is boarded, you cannot board again. So this is among all the stations. So we have one system running for all of our customers. So we started with one customer, then later on became a routine process. Whenever we sign the sales team, it’s all enterprise customers. When we sign the transportation group, then we have a routine. Basically, in two weeks, we’ll bring all of the stuff online, and before they have all kinds of a problem and they, they’re totally new to e-ticketing, they have no IT team to do that, and have some of the customers put on their websites, and they get hacked all the time. And so when we do that takeover using mainframe support, this was a Z12 on the Lennox one, and we get rid of all those headaches. It’s safe, reliable, and we only have one system to support it. We have a very small IT team, but it also runs very, very smooth and very well.
Steven Dickens: Okay. So you moved from the online sort of travel world into a ticketing space for buses. Tell me about what you’re doing at the Weather Corporation. If you can maybe spend a little bit of time, understand that maybe pre-launch. So share what you can with the listeners here, but really interested to know what you guys are doing.
Shuang Chen: Yeah, we can, at this time, share you with a high level, and basically, it’s that we use the historical weather data on real-time science data, a large amount of, thousand of sets. And trend using deep learning trend and neural network models, which can basically off the trend, that becomes, you can imagine, that becomes a smart meteorologist. And they can constantly predict based on the weather change; they can predict specific airspace and specific weather parameters 10 or 15 days ahead, just like weather forecasting. By doing this, we can accurately model, any model, the entire airspace from the weather point of view, an entire model that for any given region. So combined with another technology, the founder of this company has been working on it for the past 15 to 18 years to prove the markets. That technology basically uses a charger particle way and ionization way to convert the moisture concentration into the water, fresh water.
So when mother nature is not ready to rain, actually mother nature, as we observed every day, there is a 90% of chance when you have a moisture concentration, you don’t rain. It’s only less than a 10% chance when mother nature lines up with things together, temperature, pressure, wind, and they’re going to drop rain or snow. So Weather Corporations’ new solution is using this fully leveraged AI, and we established the detailed weather model so that we can combine those remaking technologies so we can intentionally leverage whatever moisture concentration is blown over your space. We can have a strategy to make the rain, make it fresh water. Remember, fresh water, it’s something you cannot dig from the ground, and you cannot make it in a factory. It’s nature. So the freshwater we got today from rain, it’s the same water from the last ice age rain; the last dinosaur was about 120,000 years ago. That’s the same water. You don’t get more. You don’t get less. Yeah.
Steven Dickens: Fantastic. Sounds fascinating. So Shuang, I always ask a couple of questions on the show, and I’m super fascinated to get your answers. You sound like you’ve had a really eclectic and sort of interesting career over the mainframe. So maybe let’s start to look ahead. You’ve got a crystal ball; you can start to look maybe three to five years out. Where do you see the mainframe in that time horizon? Where do you see things going?
Shuang Chen: Yeah, I’ll just give the perspective from my own point of view. I could be wrong. But with my past 20 years on the different mainframes, we do have a good sense of the past and the corporate priority and the strategy of IBM. So what I see is that mainframe probably is the only left mainframe computing system. It has a very solid integrity, you know, a high-integrity computing system. It will continue to play the backbone of many of aspects of the economy. Just consumers don’t see it. Ordinary people, you don’t see it, and you call 911, and you get a response. Many people don’t know, even 911 was… Like in Los Angeles when you’re… 911 has been running on TPF for many years. And there are many, many examples of that. So, the mainframe, in a few years from now, will be an integral part of this IBM push, the so-called hybrid cloud.
So if quantum computing… By the way, IBM is the first commercial quantum computer seen in the field now, right? You’re seeing you start and running. And this year, we’ll take a second generation within more than a thousand cubits, and the first one, it’s about 430 to 32 cubits, is the osprey, so it’s already in the customer sites. I know there’s a pipeline there, and there will be more than, of course, more than one customer site. So with this hybrid cloud platform and framework, you have anywhere from quantum computing to the mainframe, to age computing, and from private to public cloud. So I think those continuing to play a critical role in other systems are very hard to replace.
Steven Dickens: Yeah, I would agree. The other question I ask of all the guests, and I’m particularly interested in your answer to this, is what advice would you give the younger Shuang as you were coming out of college? So you have the opportunity to go back and speak to yourself at age 21, 22, but with the experience of hindsight. What would you be saying? What would your advice be?
Shuang Chen: That would be everybody has their own life path, and there’s so many exciting things to do right in our lifetime in the world. And if you ask me, that will be very biased from my view and past. AI, if you have any interest in that, get on AI. Take a…
Steven Dickens: Very good advice for the time!
Shuang Chen: Yeah, it’s not just because of a buzzword, and I told you my first message was on AI 40 years ago. So you can say I’m biased, but if you see from the eighties to now, you don’t have to…
Steven Dickens: You were just a trendsetter, Shuang; that’s all. You were a trendsetter early in your career.
Shuang Chen: Yeah, you don’t have to see very far because the CPU and computing infrastructure has reached a critical point the software can run fast enough. Now you have a better algorithm called modeling, right, the deep learning. And so this was AI based on mostly. Now you can handle a huge amount of data, an ocean amount of data, at a fast speed. You can have so many ways you can change life. We’re working on the weather. The climate was definitely one aspect, but there are so many ways. And if they get off college and they can catch up quickly within a few years, there will be, you take the right of this wave, you will do something very, very exciting.
Steven Dickens: Yeah, I’d agree. I think that’s a great way to wrap here. Shuang, I really enjoyed the conversation. Everybody, you’ve been listening to the I am a Mainframer podcast. I’m Steven Dickens. Please click and subscribe. We’d love to have you connected to our community here, and we’ll see you next time. Thank you very much for listening.
Announcer: Thank you for tuning in to I am a Mainframer. Like what you heard? Subscribe to get every episode, or watch us online at openmainframeproject.org. Until next time, this is the I am a Mainframer podcast, insights for today’s mainframe professionals.
