Connor Krukosky Talks About His Unconventional Path into Mainframes and IBM
Reg Harbeck talks with Connor Krukosky, young mainframe enthusiast, about how he got into mainframes and what he loves about the IBM Z platform. Listen to the interview via the orange play button or read the transcript below.
Reg Harbeck: Hi. This is Reg Harbeck and today I have the honor and pleasure of being here with Connor Krukosky, who is a brand new mainframer who found his way into the mainframe platform by buying a used mainframe and setting up in his parents’ basement. He has been working on the mainframe for IBM now for a while. Connor, welcome. Maybe if you could just give us a bit of background about yourself and how you got to be a mainframer at IBM.
Connor Krukosky: Thanks, Reg. Yeah, so I mean this all started—I don’t know—about three years ago now, I guess. You know, before that, I was—I had the hobby of collecting older computers. I just kind of stumbled across it as a hobby that people did and I thought it was fun. It was interesting to learn about all of these machines that competed and never made it. You know, one of the machines that I learned about was the mainframe and I learned that while it did make it, I just never knew that it still existed. Everybody talks about it in an old fashioned so, but I learned that they still make them and I just happened to stumble across the one that I purchased. I used that as a kind of stepping stone to learn about the machines and kind of just have fun with it. You know, it was just another project to me. I didn’t see it really taking me anywhere at that time.
Harbeck: So you sort of accidentally stumbled across the pot of gold at the end of the rainbow when you got that secondhand mainframe. I understand that it was a bit of a challenge to get it set up in your parents’ basement.
Krukosky: Yeah. It was a bit of a challenge to get it to fit in the first place because my parents don’t quite have a walk-in basement. So I had to disassemble it all the way down to the frame, which in turn made it easier to move as well. These things weigh a lot, but in doing that I was able to kind of learn about how it was assembled and what kind of components made it up. I eventually got all the pieces down there and put it all back together. Somehow it worked even though it went through all of that. It kind of goes to show the durability of these machines taking them apart. I mean, it wasn’t even—I didn’t statically protect anything. I just kind of threw everything wherever I could throw it and it still worked after all of that. It just kind of goes to show how well-built these things are.
Harbeck: Yeah. Now looking at the picture of you, it looks like the particular mainframe you got although as mainframes go, is not particularly large. It looks like it is sort of about what? Seven foot tall and about the same look as a large black refrigerator. Is that a good way to depict it?
Krukosky: Yeah, large black refrigerator that is about twice as deep as a normal refrigerator. This is a business class machine that I got, which, business class being the smaller-sized machine. It is half the size of the high-ends so if you look at the z14 for example, it is two racks. Mine is one of them, for example, so it’s half the machine. It’s maybe a quarter of the resources of what you could have gotten on the high end. It’s a lot lower-end machine but it’s—I’d say the front is about the size of a refrigerator but it goes a lot deeper. It’s about five or six feet deep, so it’s a big. It’s a substantially large rack but it doesn’t make anything cold, that’s for sure.
Harbeck: Yeah. It is not water-cooled I guess?
Krukosky: Ah, no. Mine is completely air-cooled but—and people think, “Oh, wow. That must heat up your house real quick,” and that’s what I thought it was going to do, but then I thought about it. It consumes about 2.2 kilowatts, which is about the consumption of a small ceramic heater.
Harbeck: Wow. So it actually uses less energy than a fridge.
Krukosky: Umm depending, yeah. 2.2 kilowatts. Now it does that all that time, consumes that period. If you power it on even if you are doing nothing or if you are running full load, it’s going to consume the same amount of power.
Harbeck: OK.
Krukosky: So that’s one misconception some people have, they think if their workloads, if they keep their workload down, they’re going to consume less power. That’s not quite how it works.
Harbeck: That’s like deleting a lot of files to make your laptop lighter, eh?
Krukosky: Yeah.
Harbeck: Now you immediately got this thing up. Well, I guess “immediate” is too strong a word, but you soon got this thing up and running and actually running some software on it. Was Linux the first thing you got on it?
Krukosky: Yeah. So, I was kind of looking around trying to find something like z/OS or z/VM. I knew of some sketchy copies of z/OS and stuff but I wasn’t really sure if anything would work. I just figured Linux is going to be easy to get. It’s the kind of legal thing to do and whatnot and I’m familiar with Linux. At that point, I’d never touched z/VM; I never touched z/OS so I figured instead of, I’m already beating my head on learning a new machine. I may as well stick with an operating system I’m familiar with.
Harbeck: Makes sense.
Krukosky: So and I went to SUSE’s site. I signed up and I downloaded a copy. Of course a month later I got emails about how my trial was going but they’re trying to sell me SUSE for my mainframe but I don’t think they realized the situation. Later at SHARE, I ended up talking to some people from SUSE. We had a laugh about it, but yeah. I was able to quickly get it up and get the installer IPL’ed. I was able to get that loaded but I had no storage. That was the next big kind of issue.
Harbeck: Now how much memory did this mainframe have? I know we traditional mainframer’s treat memory as storage and then disk as DASD. I am assuming when you say storage, you are thinking of DASD, but how much memory did you have?
Krukosky: There’s eight gigs in my machine.
Harbeck: That’s not bad.
Krukosky: No, it’s not bad when you think about, you know, it came from about 2004, but even nowadays eight gigs is OK, but it isn’t that quick of a machine. From what I could find online, I think it is about an 800-megahertz processor and that’s without being a lower end model where it is knee capped. Mine is a model 320, meaning that it’s got three out of the four cores you could have available and it’s two out of the seven speed that you could have so mine is right down there towards the end with knee capping. It’s very slow. It would probably be equivalent to maybe some older Pentium IIIs. It was kind of slow for the day but it was a lower end model. It was for people who didn’t need much and its application was at a university so it was only for education so, it made sense. Of course, it is nice that you have the options. The fact that that machine probably—I think that machine maybe cost like $100,000 to $200,000 when it was new I think, to purchase it without any deals or anything of course. That was list price. Then to think you got the same technology and the same features as somebody buying the multi million-dollar model way up.
Harbeck: It’s impressive.
Krukosky: It’s cool that you could do that. Yeah.
Harbeck: Now fast-forwarding from there, you got the thing up and running and then you came over to SHARE in San Antonio. You presented and IBM sort of took you under their wing and brought you over to Marist and Poughkeepsie and all these wonderful places. Now, I see you recently bought a house I guess in the Poughkeepsie area. Do you have your mainframe in your new house?
Krukosky: Not yet. I’m working on it. I’ve got a lot of stuff back at my parents’ house.
Harbeck: Don’t we all?
Krukosky: Yeah. The mainframe is kind of that bottom of the list because it is behind shelves of other stuff.
Harbeck: Do you still dial into it?
Krukosky: No. It’s not running because it cost about $300 a month in electric. Yeah. It gets a little tiring to pay for that. Electric is a little cheaper up here in New York so we’ll see.
Harbeck: Cool. Go ahead.
Krukosky: I was going to say I think the electric is about twice as much down in Maryland as it is up here so, significant difference.
Harbeck: Now besides that, I guess you probably got access to all kinds of current images on IBM’s mainframes in Poughkeepsie and maybe some of the ones in Marist. What sort of stuff have you been playing with on the mainframe recently?
Krukosky: When I had some machines set up and I have a little bit of time, it’s just fun to play around, newer Linux instances and at one point I was doing a little bit of performance testing for Linux on z and the newer machines are, you know, amazingly faster. The old-my machine, being an 800-megahertz maybe running at one, two-sevenths of the speed of that, it’s dog slow compared to working on a machine with 5.2 gigahertz the current generation is now, I believe.
It’s amazing how far you know they’ve been able to push the CMOS you know process and in comparison to even other architectures. Most of them are floating around four something gigahertz as kind of their high end so the fact that we’re able to push 5.2 now is awesome.
Harbeck: And of course then there is just that whole parallelism in the operating system and everything that just really turbo charges everything.
Krukosky: Yeah, and that fact that the mainframe, the one thing it offers that no other platform offers is recovery. The fact that, tell me one other platform that you can have a spare CPU sitting there idling and have one that is running. If that one fails, it will just fail right over to the other one and completely unhindered workload. You can’t do that on other machines and then the fact that you can even do that between two physical machines. You got one of them one in a different site, one on East coast, West coast and if one goes down, the other picks up the workload without blinking. I don’t work in recovery, but it’s impressive how many levels of recovery they have. I know some people that work in that team and it’s amazing how much they test. They’ve got like, thousands of points of tests. They pull things out while it’s running. They purposely try to make things fail just to make sure that it will keep running no matter what.
Harbeck: That’s so impressive. Hey, you know, as you look to the future obviously you’re right at the beginning of what looks to be an amazing mainframe and enterprise systems career in so many different ways. Maybe if you could share some thoughts about where you see the mainframe and really high-end computing and where you see your own career going from here.
Krukosky: I think that the mainframe gets picked up for more ecommerce and I don’t know if there is any businesses oriented around, say, having a mainframe and offering ecommerce services out of that mainframe, kind of like using credit card companies. They’ll have scanners all over and that’s kind of their business is to kind of lease out that ability to transact like that, but I mean, even on websites, let’s say one of the current competitors say, PayPal or Amazon Web Services. I would like to see we’ve been trying to push cloud. I would be curious to see what happens if a company got a mainframe and started leasing out time towards more conventional uses like web hosting or a small web, ecommerce. That’s kind of where I’d be curious to see it go.
As for my own career, the year and a half almost I’ve been here at IBM, when I was here as a supplemental before I started working on the bring up of z14, I was kind of jumping around. I went to the memory team, the I/O team. I was kind of going around and learning as much as I could. Every place I went to, it was an interesting new challenge. Right now I can see myself still probably here at IBM in a few years but hopefully I will dig a little further in development. Right now, I’m kind of getting on the development team and it’s fun new challenges with hardware and stuff that, we’re just talking about at this point. I like that part of the development process, working on something that doesn’t exist yet and bringing something to life from paper.
Harbeck: Cool. Well, you know, I sense that this is barely the beginning of the conversation here. There’s just so much that you’ve sort of got ahead of you and I know speaking as a fellow mainframer, we’re so happy to have you in the ecosystem. Thank you for taking the time for this interview, Connor. Any last thoughts?
Krukosky: One last thought I would have is, for anybody out there listening, that thinks, “How do I get on the mainframe?” or, you know, “Oh, I saw the mainframe for cheap. Should I do it? Should I buy it?” The other day, I had somebody ask me, what would you tell me if I said I wanted to buy a mainframe and I said do it, have at it. Make sure you’re prepared for it and make sure you’ve got the resources, the time, and the friends to help you out, but absolutely go for it. There is more and more resources everyday to be able to learn on the mainframe. Marist offers Linux on z to go play with. You know, it’s something—just go try it. Maybe you won’t get anywhere; maybe you will. You don’t know until you try, you know.
Harbeck: I like that. Those are great closing words. Thank you very much, Connor. I really appreciate you taking the time.
Krukosky: Absolutely. Thanks.