Anna was in town for a week, and we were lucky enough to convince her to stop by for a quick chat about zHyperLink, mainframe performance, and what it’s like being a globetrotting mainframer. For the full audio experience, make sure to check out the Terminal Talk podcast on iTunes and everywhere else. 

Transcript below is an approximation and may not accurately reflect spoken audio:

Frank:                    Welcome to Terminal Talk, a podcast about mainframe and mainframe related topics. I’m Frank.

Jeff:                       I’m Jeff.

Frank:                    And we are very lucky to have a top tier subscriber with us on…

Jeff:                       Terminal Talk Top Tier subscriber.

Frank:                    At the cost of thousands of dollars.

Jeff:                       Worth every penny.

Frank:                    Exactly, Anna

Anna:                     Bonjour.

Frank:                    Look at that, wow. I think she said hi.

Jeff:                       Welcome to the Terminal Talk studios in Poughkeepsie and the US in general.

Anna:                     Well thank you very much. It’s been really an honor for me to be here. I have listened, you know, to practically all of your podcast.  Probably I’m your biggest fan.

Jeff:                       Absolutely.

Frank:                    Yes, I don’t think we’ve listened to all the sets.

Jeff:                       There have been ones I have, uh…

Frank:                    Missed?

Jeff:                       Taken shortcuts.  Nobody cursed or anything.

Frank:                    There you go.

Jeff:                       Good on you.

Anna:                     Thank you.  This is great.

Frank:                    So Anna, what do you do beside listen to Terminal Talk?

Anna:                     That’s a good question.  So I’m a Z-champion.  I think that says it all. So I’m a mainframe technical specialist.

Jeff:                       So as a specialist, what do you specialize in?

Anna:                     So I support customers with the mainframe project. So I think I have quite an interesting history of performance with IBM.  I started seven years ago in Russia as a ZCTS, Z-Client Technical Specialist and I was aligned to very big accounts.  At some point, I decided that actually it would be shame not explore, you know, the world given that I work for international company.

So I was looking for other capabilities and other opportunities. And I moved to the UK. And I had fantastic almost three years in the UK. I was aligned to financial customers and it was a great opportunity to actually see how mainframe customers work in production.  Like, it’s real production 24/7. And, you know, sometimes IBM likes to say oh, here’s a CTS for you dear customer, can you install this night? And the customer says yes, our next planned IPL is in two months or something like this. So it was a really great opportunity and while I was in UK, I was introduced to the large system performance.

I became for the most part an intern of worldwide performance team.  So performance grew.  Martin Packer and Dave Bettin are my mentors.  So thanks to them I’m exploring this world. It’s really amazing because performance absolutely is a great, interesting topic. And for me it’s been a great experience.

Frank:                    Where did you present?

Anna:                     So we are regular presenting at technical universities in Europe since we’re in Europe. And a GC conference is done this way in the UK. And actually after fantastic two years here in the UK, I moved to IBM in South France. And here I’m focusing mostly on the blockchain. But this is interesting because I will say one more sentence about blockchain, then I’ll stop.  Because it changed the perception of the platform, because you know, people used to think that, you know, mainframe is a cobble machine, right?  And now it’s not a cobble machine. Now mainframe is a strategic platform for blockchain. So, can you see the difference, right?

Frank:                    Yes.

Anna:                     So I think it’s important. And also as a Z Champion, I’m leading a workgroup focused on zHyperLink. So this hyperlink is a new I/O technology.  So actually I have this great chance, you know, to be sort of like, you know, a gateway between customers and field, because I work in the field. I work with the customer projects. And communicate with the lab because, you know, it’s important to let customers – to let them know about all this fascinating technologies that are coming out. And now to also send their feedback, back to the lab so, you know, labs can analyze and do something differently.  So I had lunch with (Harry) and a friend yesterday, so I think I’m all set about it now.

Jeff:                       So you also talked a little bit about the fact that you’re doing stuff in performance.

Anna:                     Yes.

Jeff:                       What moved you into performance?

Anna:                     So actually I had some benchmark projects. And actually my first business trip with IBM was to Poughkeepsie.  So, can you believe it? Me, I was 22 and it was my like first big job.  And three months later they tell me, oh, you’re going to go to New York as your first business trip. And it thought well, whoa, this is a good start. So…

Jeff:                       And you’re thinking Manhattan?  Then no, Poughkeepsie.

Frank:                    It’s sort of like Manhattan.

Jeff:                       Yes, sort of.

Anna:                     I mean it’s still New York, right?  So I could tell some friends I’m going to New York, like yes. And so I was introduced to benchmark.

Frank:                    Utica

Frank:                    That’s right. It could be worse places. There’s worse.

Anna:                     Yes.

Jeff:                       Sorry to the one Utica listener.

Frank:                    Sorry.

Jeff:                       Carry on.

Anna:                     Oh, please, I don’t want to interrupt.

Jeff:                       Somebody has to.

Anna:                     Also, actually, we were lucky in IBM Russia, we had our own mainframe so we could do the PLT ourselves. So pretty cool, we could do POC and like three benchmark because all this we didn’t have the tools and stuff.  But at least we could test the platforms and let others migrate their solutions. And so I went to Poughkeepsie with a benchmark project. And I remember I was here for three weeks and I remember the benchmark room. That’s all I remember because we were actually very busy. We had very complicated case that we were working nights and days, but it was fantastic from the skills perspective of (unintelligible).

I worked with a performance benchmark team. So I learned a lot. And this is how I first, you know, got like a feeling of this. And then when I was in the UK, I worked on a case. And then Martin Packer was doing a study, performance study for this client.  And I remember how customer was looking and listen to Martin. I thought well, actually performance is fantastic. People get to listen to you like this. I was to do the same.  And it’s quite intriguing, it’s like interesting.  You’re solving a case, right? And you wondering okay, I will take this thread and where it will lead me. And it’s really interesting.

Jeff:                       Nice.

Anna:                     Yes.

Frank:                    Do you happen to remember, like, what – if you’re ever going back to the crime scene analogy, what was, like, the murder weapon – what was, like, the conclusion?

Jeff:                       What was the area you majored in on?

Anna:                     I’m trying to do the MBS performance.  So currently my assignment is, you know, to map a CMS113, hardware data. So that’s, you know, the team tool can analyze the data. It’s fantastic. And you can get a lot of information from there.

Frank:                    So the thing you’re working on right now that we’re going to talk about is zHyperLink

Anna:                     zHyperLink

Jeff:                       They hyperlink.  So what the heck is zHyperLink?

Anna:                     So, you know, Z-14 was released, I think there was a big focus on pervasive encryption.  And I think many people forgot that actually it was not the only technology that was introduced with Z14, right?  So it wasn’t hyperlink.

The hyperlink it’s a new technology. And, you know, sometimes you might want to ask yourself is there a way to, you know, to cut the response, the application response time by 50% without changing the application?  Is there a way to bring the data closer to the processor?  And guess what the answer?

Frank:                    There is now.

Jeff:                       zHyperLink

Anna:                     Yes, bingo.  Yes, so zHyperLink is a blend of technologies. It’s the enhancements on part of the mainframe on storage because this is really tied cooperation between storage and mainframe basically. It’s the link. It’s the direct link which is attached to the DASD.  You ask that system directly to Z-14 I/O and if you are eligible, you will see the benefit from DB2 performance improvement.

So actually I have a performance claims which I’m just going to read them. So zHyperLink improves application of response time by up to 50%, right? And it’s designed for up to 10 times less latency than high performance FICON.  These are the claims, the performance claim. And obviously customers they might see different results in their environment because it really depends on the workload.

So if you are a DB2 customer, you should be really looking at zHyperLink because really no application changes are required. All you need to do is really place your system closer to the mainframe while – because here are laws of physician which unfortunately we cannot bypass. So you have to be no longer than 150 meters. Obviously, this is the results that we have seen in the lab, right, because that zHyperLink is a fairly new technology. And currently customers can explore DB2, right?  However, there is a statement of direction saying that the support will be extended to DB2 right and this is where it becomes very interesting.

Frank:                    You know, any time you see performance numbers like that, you know, you just, kind of, imagine them a mile of asterisks behind them.  You know, there’s small, fine claims. However, those numbers area so interesting and compelling and, you know, if you are familiar with the platform, you know, kind of, what we do by shortening pathways and providing direct links.  I think that those are the types of things that somebody might start rearchitecting their application around or their environment around to take advantage of them.

Anna:                     Well this is the thing, like you don’t really need to change your application. You just simply, you know, move your storage closer to the mainframe.

Jeff:                       Oh, okay.

Frank:                    Move it right over.  Yes, we’ll start pushing it, yes.

Jeff:                       So it is actually dependent on the physical distance.

Anna:                     Yes.

Jeff:                       Because we haven’t been able to speed up light yet.

Anna:                     Yes, so the thing is that you’re probably familiar with the I/O processing in the mainframe. So, you know, that there are two types of I/O. First one is coupling facility link which are very fast, right?  And second is the traditional FICON.  So what happens with the FICON is actually, you know, whenever there is a requirement for I/O, the (unintelligible), right? And your task is already fetched.  So this HyperLink paradigm is actually – comes from a complex facility. Because complex facility link they are so fast that actually there is no (unintelligible) and this is the same what happens with the HyperLink.

So this is where the reduction of time of connect time is coming from, is because there is not (unintelligible) of the task.  And I want to underline that now the HyperLink doesn’t surplice FICON high performance. FICON is complemented, right. So, please.

Jeff:                       They’re taking away our FICON.

Frank:                    No.

Anna:                     No, your FICON…

Frank:                    Put your tweets away.

Anna:                     Yes, so this is the enhancement on top of FICON and there are some, of course, because currently it’s against DB2, but only but the support for other workloads will be growing. So if you’re planning your environment now, think about the HyperLink. Think about preparing environment. And what our performance team is – would like to do is actually help clients quantify the benefit, right?

So, you know, having the – seen before hyperlink and after and to be able to compare the results, this is something we’re looking at. So if somebody is willing to send some data, then we can probably take a look at it.

Frank:                    So it’s not worth trying to think about it as part of the com seq or the typical I/O structure. It’s something else, kind of, to plan around.

Anna:                     Well, it’s still part of the I/O part…

Jeff:                       Systems.

Anna:                     Yes, I/O system, however, you know, it’s a new card (TCPI) press card, small – so this is where the difference comes from.

Jeff:                       The numbers, you know, the 50 times speed up and the sometimes speed up, that’s – if there’s, you know, if you give DB2 users the opportunity to write a Christmas list with one item, it’s I want things faster and with less latency. So this must make them very happy.

Anna:                     Right, so the whole idea was, you know, to reduce the latency because if you take a look at the breakout of now of typical transaction, you will see that there is a hunk of time which – where the transaction is spent doing nothing. Just waiting for the data. So it’s the connect time. It’s the connect time component of the I/O and this HyperLink is aimed to reduce this latency.

Frank:                    So we were talking about DB2 because you have numbers for them, right? But it doesn’t’ have to be DB2, any kind of read would be affected, right?

Anna:                     Well, so now it’s DB2 read, right? And again, it’s very new technology, right? And, I mean, 90% of mainframe customers, they’re running between their shop. So they will the first one who will be able to explore this. However you have CICS for example. And if CICSuses DB2, then you can also explore it, this HyperLink by DB2.

Frank:                    If I have a CICS application and it’s leveraged VSAM, would I be able to get?

Anna:                     Well, so there is a statement of direction about upcoming support of VSAM. Unfortunately, I cannot say more than this. So get – be ready.  VSAM support is coming. So get read.  And what I can tell about is actually the fact that if you subscribe to, you know, to IBM, that notification. We have seen that a new function has been released, and it can dynamically switch on and on HyperLinks. And this is pretty cool. It’s done on the DSMS storage plus, right? So you define storage plus for zHyperLink and then you’re on 15 minutes without zHyperLink. And then you dynamically change to use zHyperLink for another 15 minutes. And then you can do the conversion. So this is very interesting.  So the support is very granular.

Jeff:                       At the dataset level?

Anna:                     Yes, and actually SMS has been enhanced to incorporate these changes. So there were changes to 74 subset 9. It’s synchronous I/O data information. And to 42 subset 5 and 6, which are represent storage class and dataset level. So and if you’re using Z-(unintelligible) you can do some preliminary performance estimations using the (unintelligible). If you want to do like the real performance studies, then contact our performance team.

Frank:                    Yes, so how much time do you spend doing a lot of this like, kind of, lab performance work? And how much time do you spend out?

Anna:                     So, actually I’m really dreaming about, you know, doing performance fulltime. So I think I have expressed this a couple of times to some people.  But I at the same time I really like the customer facing role, because customer asks very different questions.  And motivates us to do something definitely to, you know, or to have their feedback. It’s very important because in the end we’re creating all this (unintelligible) technology for them, right?  So I like this combination of lab work and still customer visiting. However, I would be really keen to do more performance work. Wink, wink.

Jeff:                       Well the worst guarded secrets.  What can you tell as a world traveler, literally a world traveler? What can you tell us about like who mainframe customers are around the world? Are they all, you know, can you walk into somebody in France and, you know, is it all the same or is there different things you have to keep in mind?

Anna:                     Well, I would say that mainframe customers, they are the same and different at the same time, because I would say that they have one thing in common. They are very smart and intelligent, you know. And they’re very experienced. So you need to come prepared.

Jeff:                       I think you’re just trying to pump up the listeners here. I mean obviously mainframe users are some of the smartest, best looking.

Frank:                    Most, let’s say most, not all.

Jeff:                       But, you know, so they’re smart, but, you know, is there – are there a difference in market? You know, obviously, you know, DBPR and everything is …

Frank:                    Yes, well obviously, you know, they have to meet different types of regulations because, of course, all of them are processing sensitive corporate information. So they’re really concerned about security. So, and to – they differentiate, in a way, how they’re regulated, what’s the workload they’re running, and, you know, some customers, they like to run very hot.  You know, like very close to the limit of the mainframe because well, you know, it’s what mainframe does, right? It can run pretty close.  Some people like to have some additional headroom.

So customers, they really have something in common at the same time, something different because, you know, it’s not them who are deciding how (unintelligible).

Frank:                    Right. You said that you – a lot of what you’ve been working with from a customer perspective is financial. Are there other areas of the industries that (unintelligible) or…

Anna:                     Yes, so I think I had this unique experience of working both with enterprise customers and startups and young companies, predominantly for the blockchain project. But it’s a great experience. And on the one hand, we have enterprise defied. And we know how they work, and at the same time, it’s very challenging to talk mainframe to young companies, to startups. But I think it’s a great thing to do because we need to make sure that all people are using our (unintelligible) technology.

Frank:                    Yes, so let’s talk a little bit about that then. You said that you’d been talking to these startups. Are they asking a different set of questions from the enterprise client?

Anna:                     Yes, they do because the thing about startups is that they want to get up and running very fast, right? And they want to scale out very quickly. So this was their concern, and also, they would really like to know how we support different type of workloads, how we support Docker and containers into (unintelligible).

Frank:                    Most of those clients that you’ve been working with in that space, have mostly been in the Linux side?

Anna:                     Yes, the part of the Linus workload is growing.  However, I think that it is important to equally communicate the value of z/OS, right because this is the system, right, to build most reliable your solutions. And I think that we should not, you know, hold ourselves from talking z/OS to startups, because why now? It can be like, you know, a startup will become a bank in a year. So they will be running the transactional workload. So why not?

Jeff:                       I would love to be able to present z/OS as something that you can graduate up to.  You know.

Frank:                    And I think that’s why it’s really important that we get the whole Dev Ops things working because if we make it easy, do you develop to?  Especially if you’re startups, right, it would be a lot easier for them to do that gradually.

Jeff:                       Yes, it’s just a container. I’m going to throw my thing on that, you know, on that platform.

Frank:                    Just so happens to be very powerful.  We talk a lot about containers. Are these startups really focusing on Docker a lot would you say?

Anna:                     Well Docker is the de facto like a standard in the world of containers. However there are other different, you know, solutions. So I think everybody is willing to know what we are supporting, right? And actually I’m happy to say this. We’re supporting lots of stuff. And like three years ago, there was a chart what Linux wants support, right? And it was just like a single chart with some small logos of the company. And like, you know, now we have like a deck where we can show all the solutions that we’re supporting on Linux one platform. And this is great. This is a great transition.

Frank:                    You started in Russia.  And you’ve done Europe and obviously a couple stints here in the US.  Did you start as a teenager saying I want to work on mainframes? Or how did you get to mainframes form?

Anna:                     So I have a funny story to tell about how I got into mainframes.  So oi was a PhD student. And like all PhD students, I was moonlighting in the research center.  And at some point, these research center, they bought the mainframe and they said if somebody’s willing to work on the mainframe, we’re going to pay more to this person.  I was the first one. And I spent like a year to Z-OS and DB2 for the installation, SAP, and all this, you know, like (unintelligible) stuff.  Actually it was a very good school.

And then I got into IBM as an intern, and I got into the systems architect team. So I was supposed to be an architect like, you know, like George Costanza.

Jeff:                       Oh yes, nothing against architects.

Anna:                     Yes, and on my first day as an intern, I was sitting in the lobby and I was waiting for my manager, you know, to badge me in, to give me the laptop and all that stuff. And then I saw mainframe team. And they were like and they knew me from – because they were coming to this research center.  And they were like well why are you sitting here? You can wait for you manager in our room. And the thing is back then like everybody sat in open space except the mainframe and storage team. They had their own room. And you had to badge to go into this room, and this room was just in front of the data center where the mainframe was. So I sat there.  And I never left.

Frank:                    So you saw it as an elite thing not like put those guys in a cage, you know.

Anna:                     No, I mean like, you know, other people could not badge into this room. It was our private room.  So even sellers could not go into this room. So yes, I think that what I like about mainframe is the scale. Because I have this chance to work with a big clients and with the new, exciting startups. Travel all over the world. I’ve worked in Russia, in UK, in France. And I really like the scale because there are some things that you can do only with the mainframe, right? And this is very exciting.

Frank:                    What was it like to get started on a mainframe when you didn’t have a ton of people supporting you?

Anna:                     So when I started, I had like this pile of IBM documentation. And since I graduated with engineering degree, all I need to have is a manual, right? And I can figure this out. So I know that some people were like, you know, scared. Oh no, there is IPSF green terminal. I actually found it quite convenient once you master it. Once you learn how to use it. It becomes very convenient, very fast.  So I didn’t have any formal mainframe education until I started presenting on the mainframe.

Jeff:                       And then you just, kind of, if you needed to figure something out, you went back to the lab. Learned it. And came back out.

Anna:                     Yes, and I mean when I’m really grateful for is the ecosystem that we have. I mean there are so many great people that I can ask the advice if I’m stuck. So I never had any issue of getting any question solved. So this is fantastic to have this ecosystem. Like, you know, (unintelligible) center, (unintelligible) center obviously Poughkeepsie and this is great.

Frank:                    So how long after you started with this stack of manuals?  You guys missed it because you put her hand about four feet from the floor, and the stack of manuals. How long did it take you to feel comfortable managing the system?

Anna:                     Well, let me reply. So there were a couple of (unintelligible) and actually they were the fastest learning curve.  You know, when you’re breaking things and when you’re fixing things and you’re learning from this, so that was really helpful.

Jeff:                       You can really – in a timeline of one’s career on the mainframe, you can really set those flags, you know, mileposts of when you really screwed something up.

Frank:                    It’s the best way to learn.  You never learn when things are going well, right?  Awesome. Well this is, I think we’re at the bottom of the hour, right?

Jeff:                       Yes, we are.

Frank:                    So I really want to thank you, Anna, for coming and being part of this and telling us about your experience which I think is pretty unique among the people we’ve interviewed.

Anna:                     Thank you. It’s been a great honor for me. And I really enjoyed. Thank you so much.

Jeff:                       Thank you.

Frank:                    Old man Charlie, run us out.