Skip to main content

Martin Packer on His Mainframe Performance Journey

Reg Harbeck: Hi, this is Reg Harbeck and I’m here today with Martin Packer who is a mainframer based in the United Kingdom (U.K.) and among many other things works on performance, works with Marna Walle in a mainframe performance podcast and has been quite active in mainframe in a lot of different ways. Well Martin, why don’t I let you introduce yourself to us and tell us, how did you end up on the mainframe?

Martin Packer: Well I guess it was inevitable because my father was in the Royal Airforce and teaching computing in the Royal Airforce in the 1960s and into the 1970s so there were a lot of computer books on the shelf at home including the old Elsevier Computer Monographs which I lapped up and if we scroll forward a bit, I was doing a master’s degree in information technology at University College London in the mid 1980s, got myself a Z80 based computer called an Amstrad CPC 464. Yes, I did program it in Assembler and gasp, actually in machine code hand planted. Then I joined IBM in 1985 as a young systems engineer. I went through the training for the 18 months and became a large systems specialist.

Reg: Okay. Now how early on did you find yourself working in performance?

Martin: Well it’s one of these things that creeps up on you because I was doing some stuff in the branch as a large systems specialist which wasn’t heavily performance related but had some things to do with virtual storage which was hot then and I guess slightly more tepid now but still interesting for a bunch of customers. Then I moved onto the Lloyd’s Bank account team and then ended up doing some things that supported data in memory which obviously were much more related to actual performance. Then in 1990 I moved to a country role doing performance and capacity so it gradually crept up on me.

Reg: Now of course there are a lot of different kinds of performance. There’s real-time performance. There’s the ability to automate based on performance. There’s capacity planning and performance monitoring. There’s using SMF. What sort of areas in performance did you start in and cover during your career?

Martin: Well I like to explain that what I do is basically nosiness with SMF so I count as a couple of flavors of that question. One is generally primarily my evidence or instrumentation is SMF a customer sends me though it’s not exclusively SMF but the other thing is whatever you can get out of SMF that might be interesting or useful is what I want, not just performance, but nosiness about architecture and how people run things, when they stop and start their CICS regions, for example, which Db2s the CICSes attach to. This has got a little bit to do with performance but it’s a much broader “wringing the hell out of the SMF data” kind of approach.

Reg: Cool. Now one of my first jobs as a young CICS system programmer at the beginning of my career was actually to write a C program, if you can believe it, to take apart SMF records and get some information out of them. That was a really mind-bending experience as you might imagine, using C to figure out dynamically how long is a record and all that kind of stuff. I’m going to guess you probably use slightly different tools. What is your favorite tool to take apart and look at the content of an SMF record?

Martin: Well we have for many years been running with something called service level reporter (SLR) which actually went out of service I don’t know how many years ago—let’s say 20 years ago. But there’s an awful lot of code that was built by the giants whose shoulders we stand on in SLR. It’s a nice query language and you can drive it with Rexx and you can drive it to produce GDDM charts. More recently though, and actually I’ve got a blog post about this, I’ve been using Rexx to process SMF because now you can.

Reg: Oh, directly? You can use Rexx?

Martin: Well a feature of z/OS 2.1—which I had to explain to Marna what the utility of it was—is the ability to process VBS records and SMF is VBS records so now yes you absolutely can. For low to medium volume data, probably not CICS monitor trace to be honest, Rexx is ideally suited.

Reg: That is so cool. My favorite programming language is Rexx and of course one of my personal heroes happens to live, you might say, driving distance from you, the fellow who invented Rexx. Now I’ve never talked to him in person. I’ve exchanged emails with him so I’m going to probably mispronounce his name. Is “Co-lish-ah” or “Cow-lish-aw” because I would say “Cow-lish-aw?”

Martin: I would say “Cow-lish-aw.”

Reg: Okay. So actually we’re just writing an article about him on Rexx’s 40th anniversary and so that’s going to be coming out on but now that said, I’m going to guess that you probably use Rexx for a lot of other performance-related things. What would you say is your favorite thing to do with Rexx?

Martin: Mostly it’s pumping out CSVs so then I can swear at Excel.

Reg: Oh, neat. You know I’ve done that a lot myself now that you mention it, just take the data on the mainframe and then just copy it across or FTP or IND$FILE. Cool.

Martin: I have to say to you it’s a bonding experience actually sitting down with a systems programmer and both of us swearing together at Excel. This has happened many times with dear friends and customers as we sat there and tried to make the damn thing do our bidding.

Reg: That is a very familiar sounding experience to me. Now that said at some point, you and Marna Walle started doing a podcast and you’re up to episode 24 is it now, talking about performance and other topics. Given that the podcast is a transatlantic podcast, I assume that you’ve rarely if ever actually been in the same place as her when you’re recording it. How did that end up happening?

Martin: Well we’ve known each other for a long time. We’ve met many, many times in fact because we go to the same conferences whether in the U.K. or in Europe or in the United States (U.S.) and I vaguely recall saying to her, “I’m thinking of doing a podcast. Are you up for it?” She said yes much to my delight and slight surprise. So then the name “Mainframe, Performance, Topics” came to us partly because I’ve actually changed my own blog’s name to “Mainframe, Performance, Topics,” a subtle change there actually because I stuck commas between “mainframe performance” and “topics” which gave us the structure so she can talk about mainframe or lead on that and I can talk about performance. We can talk about whatever else we feel like so we don’t have to talk about work necessarily as the topics topic, and that brought to us quite a nice structure to work with. We’ve been friends for a long time and we are now back to recording episode 24 which of course us being computer nerds is our 25th episode but we’ve got 24 under the belt now.

Reg: Cool. I had the opportunity to listen to your most recent one which was a preview of the latest release of z/OS, some really neat stuff there. So how did you end up in a role and more to the point what is that role that allows you to have insight into so much of what’s up and coming on the z/OS environment?

Martin: Well on the mainframe side in terms of z/OS, that’s really Marna’s bailiwick because she works in development and so has obviously privileged access. For me, I get most of my insight actually from working with customers because my day job is really a trouble shooter even though I say it is being nosy with SMF. The purpose of that is to drive conversation to the customers around the world so I get a lot of insight to things that I guess other people aren’t lucky enough to see because of my wide social set of customer contacts and involvement.

Reg: Cool. Now being based in the U.K., are your customers primarily in Great Britain or in the U.K. or are they further afield in Europe? How far away are your customers?

Martin: It started off just being a small patch in London and then it increased to being the whole of the U.K. Then it increased to being the whole of Europe and now it’s a worldwide job and has been for some number of years. Last year I was very lucky for the first time to work with a customer in Korea.

Reg: Oh, neat.

Martin: South Korea obviously and I also worked with U.S. customers last year and I’m working with customers in Denmark and visiting my customer set in South Africa.

Reg: Oh, wow.

Martin: So my customers are all over the place now and as with many people’s careers, it has kind of expanded and expanded. I don’t think there is anywhere else to go though.

Reg: Well you know I know the feeling. I like to say that I’ve been to every continent where I could find a mainframe but I haven’t been able to find any mainframes in Antarctica yet so I gather you’re in the same boat on that one.

Martin: Well I guess the cooling problem would go away, not that we have a cooling problem, but it certainly would have no problem staying cool there.

Reg: I have a feeling it would be a very Linux-friendly environment.

Martin: I would think so. I can’t imagine why.

Reg: Now that said, obviously one of the things about dealing with people is also dealing with user groups. I’m going to guess you probably have a fair amount of involvement GSE and maybe some other user groups as well.

Martin: Yes. Well I used to be involved with U.K. CMG, which is actually what you’d expect, until they stopped operating. I have been to the GSE U.K. conference many times. I do work with the capacity management and performance analysis working group of GSE. Actually they’re coconspirators with IBM for the Z Technical University in Europe every year so yes, I’ve been involved in user groups a lot. I am a firm believer of them and one day I’d like to make it SHARE—heavy hint.

Reg: Yes. We would love to see you there.

Martin: Well I think I have a treat in store whenever I get there because I have a really nice presentation that’s about to emerge.

Reg: I look forward to seeing it. Is that nice presentation you’re talking about going to be available for viewing online at some point do you suppose?

Martin: That’s my firm intention. Now it’s actually a presentation that’s being bootstrapped up from not very much at all so the question is when to strike and publish. I think the answer to those two things is publish early, publish often and it’s also publish and be damned strictly in that order but I guess if we publish the alpha version then we’ll be damned before we finished actually writing and publishing it but such is life, no safety net.

Reg: So I guess if people want to see that presentation or other stuff that you’re doing, they can go to your blog. Is that correct?

Martin: I’m sure I will be trailing large chunks of it in my blog over time and I would expect Marna and I will record an episode or two where we talk about chunks of it as well but the best place to see us typically is at the various conferences and see how far we’ve come with each iteration.

Reg: Now we live in an interesting time in the mainframe partly because we always do but partly because we live in a world where everything that’s always thrown stones at the mainframe as being old, legacy and out of date is now older than the mainframe was when they started throwing stones and is becoming out of date itself while the mainframe is still alive and well. That said, it doesn’t change the fact that the press for the mainframe isn’t that good but everything else about the mainframe is great. Given that context, what are your thoughts about the soon and longer term future of the mainframe, our viability and our role in the world economy?

Martin: Well bear in mind it’s pretty central to the world economy in oh so many ways and I would say so long as it continues to evolve and so long as we continue to invest heavily in its development—not just the hardware, but the software and the ecosystem and university programs and user groups and all that stuff—then I think it’s got a very rosy future. I would say after the 2.4 preview announcement the Docker container extension stuff is particularly exciting to see on z/OS because that’s real work load. That’s very pervasive that drives useful growth to us so I think the future is looking pretty good right now.

Reg: Well I have to say when listening to you and Marna talk about that Docker containers, I mean just the whole concept blew my mind of Linux under z/OS. I thought Linux went under z/VM. Do you know who thought of doing Linux under z/OS? Why would you do that?

Martin: Well how else are you going to Docker?

Reg: Okay, fair enough. I can’t answer that question. Now that said, we’re getting to a good point to sort of tie things up here. Any other thoughts you wanted to share, either motivational or observational or just stuff that you’d like to finish up with?

Martin: Well I think the key thing is always keep learning, always keep playing with stuff. Don’t just stick to the technology of the mainframe. Stick to the visual depiction things like things on iPads and what have you and web technologies because there’s a lot of fun to be had in this game. Actually I have a presentation for how to be a better performance specialist which is actually more a calling card for how much fun you can have doing it so I think look for the fun in the technical job. Try new things out. Find people with common cause like user groups, like IBM-Main, like MXG-L listserver and do all the social things I think are the pro tips I would give people.

Reg: Awesome.

Martin: It’s been a lot of fun so far and it’s going to carry on I hope.

Reg: That sounds like great advice and a great way to finish up the podcast. Thank you very much for taking the time Martin.

Martin: Well thank you for having me.