Jim Elliott Stays Motivated on the Mainframe
Reg Harbeck talks with Jim Elliott. The longtime, now "retired" IBMer explains why today's mainframe is leading edge technology and shares what motivates him to keep working. Listen to the interview via the orange play button or read the transcript below.
Reg: Hi, I'm Reg Harbeck and I'm here today with Jim Elliott who is a longtime mainframer and VM, Linux and VSE IBMer who's been active on the mainframe in Canada and worldwide for quite a long time. Jim, maybe if we can start if you could introduce yourself to us, and just tell us, how did you end up on the mainframe? And where have you been since you joined the mainframe?
Jim: Thanks Reg, it's a pleasure being able to speak to you about this. Well I was with IBM for 43 years until I retired in January, but I started my mainframe career, if you will, as a student at BCIT [British Columbia Institute of Technology] in Burnaby, B.C. In those days, of course we are talking 45 years ago—boy, that is scary—45 years ago basically it was mainframes and some other specialized machines. Of course, those were the days of the bunch with lots of different vendors and different architectures. So I learned in school DOS/360. I joined IBM working as a systems programmer, and I was the VM/370 and later DOS/360, then DOS/VS systems programmer in IBM's Western Canada Computing Centre. I went from there into the branch in Vancouver, then the IBM lab in Toronto working on IMS application development. From there I went down to the U.S. working in Americas Far East Headquarters, back to Canada working in the Systems Center as the country VM technical support, back to the U.S. then to launch Linux on the mainframe for the Americas, and then back to Canada, eventually ending up in 2007 as the mainframe technical presales lead for IBM Canada, which I did until I retired. Now working for GlassHouse Systems, IBM's premier server business partner in Canada. I am the lead z presales technical rep.
Reg: Cool. It reminds of a favorite joke of mine: What's the average mainframer's retirement plan? It's to come back to work doing the same job as a high priced consultant. Now as somebody who sort of—
Jim: I will say the important thing is we're working part-time most of the time. I work 20 hours a week.
Reg: Ah, nice.
Jim: That's retirement.
Reg: Very nice.
Jim: Or at least a bridge to retirement.
Reg: And probably making money that makes it worthwhile to be doing that.
Jim: Ah, yeah.
Reg: That's nice to hear. I think you know for a lot of our mainframe colleagues you know that is sort of a really nice perspective on the future is to you know look at the possibility of being able to have a genuinely positive lifestyle that involves continuing to be on the mainframe.
Jim: Yup. Staying involved with SHARE is also very much a part of that. So as a SHARE project manager/virtualization project manager in the LVM program, it's keeping me also very involved.
Reg: Excellent. Now as somebody who's sort of worked with a lot of mainframe technology, then gone through retirement and then come back on the other side of retirement, some of the technologies you've worked with have had similar journeys—especially I'm thinking of VM and VSE which, you know, I mean—maybe if you could just reflect on how did they come to be? How did they manage to avoid getting retired given the best efforts of all of those involved?
Jim: Well you know, when you look at VSE—z/VSE today—it started out as DOS/360, a program that was developed by IBM because OS/360—which was to be the operating system for System/360—they couldn't get it to work. So DOS/360 was put together as a temporary stop gap measure and it's still alive all these years later and doing very well. Then VM on the other hand started off as a research project out of the Cambridge Scientific Center, the official virtualization product for the 360/67 was called timesharing system out of Poughkeepsie. The Cambridge Center in Boston developed this as a research project. It became very popular with universities.
So we get to 1973, IBM is announcing virtual storage—yeah '73—and they needed a tool to help people migrate from OS/360 to OS/VS1 and OS/VS2. So they took VM/CMS and made it into VM/370. Interestingly enough it was supposed to have two releases and die. It was a migration tool. IBM is very glad that it didn't die now. It evolved into being the dominant email platform in the world, running PROFS for a great many years. Linux on the mainframe today would not exist if VM wasn't there to host it. And so that developed neatly. On the VSE side—by the way, you know the joke is VSE was actually formally killed by IBM at one point. There never was a VSE/XA. It went from VSE/SP, IBM resurrected it and came out with VSE/ESA. Both products are doing very well still together with the labs in Germany and in Endicott, New York.
Reg: Interesting. So IBM may have killed it and resurrected it, but I think it was the customers that were keeping it alive the entire time.
Jim: Oh, definitely. I mean especially with VM. I mean with VSE, it was a small group of very core customers who are still very interested. Here in Canada, we have some very large VSE customers—including the largest one in the world—and VM, given where the rest of the world was going, if you remember object code only, VM stayed mostly as source code. And today still much of VM is shipped as source code.
Reg: Cool.
Jim: It there became—therefore became easy for people to modify it and submit enhancements to IBM. So it got a very loyal following as a result of that.
Reg: Very cool. You know I must admit that VSE and VM are some of the reasons why I came up with one of my own personal philosophies, and that is, "the temporary outlasts the permanent."
Jim: Yeah, it's very true. But it's a common statement in IT. I have customers who are still running VAXes, DEC VAXes running VMS, even though—I won't say Compaq. HP stopped supporting it for years, but the reality is they still work. And that's one of the things about the mainframe world: You know I was doing some work with a customer a few years ago, who had some FORTRAN programs that hadn't been updated since the late 1970s, and they worked just fine on z/VM CMS. I had another customer who had some FORTRAN programs that were originally—were last complied on OS/390 2.4. Took the binaries and they ran fine on z/OS 1.13.
Reg: Wow.
Jim: You can't say that about any other platform. So mainframe things continually stay alive and evolve.
Reg: Hmmm. Now that leads me to a question I often like to ask mainframers who have been around for a while: Tell me about the oldest program or other module you've seen on a current mainframe that still exists from like the 1960s or 1970s that you've encountered.
Jim: I encountered a lot in the Y2K period, right up to 2000 with people running some incredibly old software that had been out of service for years and still worked, of course—DBOMP, which you probably never ran into in your career. It was such a niche product; it was a database. I have to say it was this one international, vertically integrated oil company who still is running some of the refineries on software that is 30+ years old, and it still works. You know they're slowly migrating to quote-unquote distributed servers, but the fact is they still have I think three refineries left in the world, scattered around the world, still running this old software that was originally written to run on VS1.
Reg: Hmm. The power of legacy.
Jim: Yeah, exactly. Legacy systems are systems that work. That's my favorite line.
Reg: Oh yeah. Now another legacy of course that you and I have dealt with a lot is the user groups and obviously our favorite one is SHARE, but there are some other ones out there as well, both local ones—I remember COUG in Toronto—and more widely used ones like WAVV, the VM and VSE one. Maybe if I could just get you to offer some reflections on past and current mainframe relevant user groups.
Jim: That's a good story. I mean SHARE being the big one in North America. Of course, there's GUIDE SHARE EUROPE.
Reg: Right.
Jim: And there is Australasian SHARE/GUIDE. The community user groups, a lot of them have gone away—like we used to have the Canada users group here in Toronto and Montreal. That is long gone, but COUG is still somewhat active and we are going to be starting it up again, hosting it at GlassHouse Systems because we have a meeting facility.
Reg: Cool.
Jim: But I can remember some of the other ones—like there was the—that are still going strong like CAVMEN out of Chicago. There's user groups around North America and around the world, but what we've started seeing for new mainframe user groups is for Linux. We are starting to see some of those evolving often out of the old VM user groups.
Reg: Okay.
Jim: But the user groups, one of the great things about it is, you see it with SHARE of course—is the community that it gives you. I mean you can go to something like EDGE, which is the IBM Server conference or STG conference, but going to something like SHARE or one of the local user groups, you actually are working with your peers, and that is very good. Some of those are gone like WAVV is gone but now we have the VM workshop, which cover VM, Linux and VSE. So this year it is in Columbus, Ohio, I believe, in June. So you know that's very good, those user groups, that interaction and that support mechanism. Sort of like that, almost like a user group is the internet lists.
Reg: Hmm. Right.
Jim: So you know IBMVM, IBM-MAIN for z/OS, VSE-L for VSE of course and LINUX-390 for Linux discussion. Those are a virtual user group in many ways. I mean the support our customers and users get out of those are extremely good these days, and they've been around you know they—when you look at IBM, they have VM involved from VM SHARE―
Reg: Cool.
Jim: Which was originally on BITNET if you go back a few decades.
Reg: Wow. So now one of the things you have sort of been hinting at that I would like to maybe get you to put on your future goggles on now is just the future of mainframe computing. Obviously Linux is a really big part of it, but it's not the only part. Maybe if I can just get you to draw the tapestry as it looks to you moving forward over the next decades.
Jim: Well it's an interesting question because we keep on hearing that—we don't hear it as much as we used to that the mainframe is dead. We used to hear that all the time. We still do hear about it with some customers where they get a new CIO who believes the world runs on Intel under Windows—not even Linux, just Windows—it's an educational thing for those executives and management by frequent flier magazines. But when you look at the mainframe business according to IBM, you look at the installed MIPS—a terrible term, “Meaningless Indicator of Processor Speed” – but it's a measure of capacity—and it's growing at about 25 percent a year. Customers are going away—some of the small ones are leaving the platform—but others are growing.
You look at where that workload growth is today, about 25-30 percent of the installed mainframe capacity of the world is running Linux. Which, you know, for something that when we released that code back in December 1999, we we’re, we didn't even know if anybody would download it. It was a research project. It turned out to be a very successful thing. The other things that are driving mainframes these days is not the traditional CICS IMS workload, it's things like analytics, Cognos on Linux and on z/OS, SPSS on Linux, all the information data warehousing stuff that runs on these platforms, things like the IBM DB2 Analytics Accelerator for doing really deep queries. Web serving—you know WebSphere runs fantastically well on both Linux on z and z/OS. Then today people think—you know I get these people who think, "CICS—well CICS means COBOL." I have customers I'm working with running z/OS right now who are actually rewriting all their CICS transactions in Java-
Reg: Wow.
Jim: And doing updates to them because you can. There is a lot of Java skill out there, more than there is COBOL skill today, and there is nothing to prevent you from writing CICS transactions in Java today. Java has—sorry—CICS has built into it the IBM Liberty WebSphere runtime engine built right into it. Then there is all the other products that IBM has got out like DB2Connect, APIConnect and WorkLight—so all of these modern tools are available for z. Some might require Linux; a lot of them don't. They run fine with z/OS. So there's all that capability that more and more is getting in there. So z is very definitely a platform for the 21st century. I don't see z going away. IBM has indicated they are going to put out a new generation every two and a half years or so. When you look at the z13, which came out—God, almost two years ago now—the z13, where you go up to 140 4-way processors with 10 terabytes of memory and they're 5 gigahertz processors. This is the fastest general-purpose processor in the world. There's incredible capacity in these things. I/O bandwidth on a z13 sustained is over 800 gigabytes a second.
Reg: Wow.
Jim: Mainframes today are very much current technology and if anything, leading edge technology. People are starting to be aware of that—I don't think—or people are very aware of it, especially in the large customers. I don't see mainframes going away. You know I don't see Power going away. These platforms will be there, because customers need them to do their work.
Reg: Cool. So we've gone a little bit over time, but I have no regrets about it. But what I'd like to do is, even so, give you another minute or so just to kind of tie up and offer any closing thoughts. This has been really good.
Jim: Oh, closing thoughts? You know, so I've been in the business now for what—43-44-45-46 years that I've been working in this stuff, because there were other companies in there when I was a student. It's been a really enjoyable career path. You know being able to work on this and continuing to work on it as long as I want. I'm not expecting to keep working forever—I'm 63 now—but at this time I'm still really enjoying it, and I think it's a really good place for young people to get into. Mainframes aren't going away. As you know with your work with zNextGen that this—we need the skills out there. Young people are interested. You look at people like Connor [Krukosky], I mean, a fantastic story. The more young people we can get interested in the mainframe, the better it will be for everyone. SHARE is a large part of that I think, supporting that growth of young people onto the platform so us old people like you and I can eventually retire. Eventually, no rush.
Reg: Sometime in the next century or two, maybe.
Jim: Yeah.
Reg: Well thank you very much, Jim. I really appreciate you taking the time for this and I really appreciate all the thoughts and insights you have shared.
Jim: No problem. It's always a pleasure talking to you, Reg.
Reg: Same here. Take care.
Jim: Okay. Bye.