This is an urgent update, and I want you to click here.
May 16, 2021

Larry Mead, Principal Program Manager at Microsoft


Walter Sweat - Astadia, CTO

00:02 Hi everyone, welcome to the latest edition of Walter’s World, the podcast series from a Astadia. Astadia is a global service provider and we focus on mainframe migrations and modernization and this podcast series is really designed to bring in some of the industry leaders in this space to help talk bout topics that might be of interest to people who are considering alternatives to their mainframe.

00:29 And today, I am delighted to have with me Larry Mead who's the principal program manager for Global Azure Engineering. Larry, thanks somuch and welcome to the podcast series.


Larry Mead- Microsoft, Principal Program Manager, Azure Global

00:42 Well, thank you, Walter and you know this excited to be here and I hope we will get some goodinformation for the listeners.


Walter Sweat - Astadia, CTO

00:50 I think it's going tobe great for people to listen to your experiences and recognize, you know the types of customers that you've worked with and the things that you've seen as you'veworked over these last many years and mainframe alternatives.

01:06So Larry, can you tell us a little bit about you know exactly what do you and your team do for Microsoft.


Larry Mead- Microsoft, Principal Program Manager, Azure Global

01:12 Well we're part of a group that's called the critical infrastructure team within azure global and what that essentially means is that we look at how to best host.

01:23 Applications that would be considered critical infrastructure, you know into azure and certainly.

01:30 Anything that's coming off a mainframe is probably critical infrastructure for the customers who are using that application so our job is to make azure.

01:40 The best place to host applications from mainframe for customers who are looking for alternatives and also very important is.

01:49 Working with partners like Astadia to have their solutions work, you know and perform in anoptimal manner within azure so Those are our main focuses, but we also work directly with customers.

02:04 You know when needed, especially the larger ones, are the ones that have maybe some things that aren't common, you know as far as.

02:11 migrations areconcerned and transformation is probably a better word to use at, and then you know once we've done that, then we have a blueprint for you know other customers if we move forward so that's essentially what our team does.


Walter Sweat - Astadia, CTO

02:27 And that certainlymakes sense, you mentioned the larger customers and Larry I know enjoy working over the years, with you on some really large joint customer opportunities.

02:36 I think the audience might be interested in your opinion, what are the two or three most importantelements for an organization to consider when they're looking at mainframetrance formation alternatives.


Larry Mead- Microsoft, Principal Program Manager, Azure Global

02:49 Well, you know that'svery good question and I would say, the first thing they should look at is what did they really need.

02:57 When they want to land in an in a different platform like a cloud platform.

03:02 By that I mean what dothey need for availability, what do they need for throughput what are their SLA is what is their Dr requirements.

03:10 What are the things that I'll call almost the non functional requirements that they're goingto have these applications.

03:16 mean the mainframes have always been pretty solid for delivering those so we have to take a look atwhat is it that you know that are.

03:25 They you know, do not want to have any compromise on as they're going into a platform like azure sothat would be the first thing.

03:35 The second thing would be okay, I am I happy with the applications that I have just because i'mconsidering moving off the mainframe doesn't mean.

03:45 that the application is not doing what I wanted to do if i'm happy with that application and i'mstill maintaining it.

03:53 Then you know that's a different kind of a transformation into the cloud than that let's sayan application, that is not maybe meeting the business needs of the customer.

04:04 Under those conditions, I might want to consider what the cloud can do for me in moving this across to maybe either enhance that application or perhaps even.

04:17 Replacing parts of itas part of that transformation, so those that's the second thing and then the third thing that is very important is how do I get from here to there because.

04:30 You know if,especially with the larger customers that's not something that's done over a week-end that's something that can take several months, it could even take ayear or more to do, especially for the ones like you and i've done that go upto like 300,000.

04:45 MIPS that's that's avery that's a very large customer.

04:47 By the way, noteveryone's that big but anything that I would say is over, say 10 to 20,000NIPs given us when we get into.

04:57 Looking at you know Ineed to have a plan for how I get from here to bear and do it in a way that does not disrupt my business in that process and i'm still my end customers areeither.

05:09 You know, being exceeding their expectations or at the very minimum, at least meeting those Soif you take care of those three things at least you're on a good path to you know do that transformation.


Walter Sweat - Astadia, CTO

05:22 I think you're spot on as always with that lay there for me it's a there are no to mainframes that areexactly alike.

05:29 Nor are there no tolanding zones that are going to be 100% the same for customers and understanding that roadmap and how you're going to get there, defining that onthe front end is critically important.

05:44 If we can let's talk for a second about past services, how do you see customers implementing past services as part of their transformation Program.


Larry Mead- Microsoft, Principal Program Manager, Azure Global

05:54 Ok, that's avery interesting topic because first off, you can do the ends with some of the transformation and there's value that you can gain from bms, but a lot ofcustomers will say Okay, if I do that well what's happening as far as.

06:12 You know I if i'mdoing it on the vm they see a vm as needing you know, a person right so.

06:22 If i'm implementingwith past services, I can probably now do things with you know, out of anotherFTE to run my stuff in the.

06:31 In the cloud sothat's, the first thing the second thing is past services, by their nature,tend to be the more elastic type services there's the ones that can allow me toscale more seamlessly.

06:45 that's true with ourdata services like sequel db you know our as your sequel db and azure or it's.

06:51 True, with our Coopernettie service aka S, so this also brings in a more seamless type ofscalability Now you can still scale with the m's i'm not saying you can't.

07:02 But that's a littlebit more where i'm turning them on and off, but, as you know, in order toresize them now, I can have.

07:11 You know where I canhave elastic pools, too, but it's it's more straightforward if you're doing itwith the past services.

07:19 And then the lastthing that I would say some of the things that's being added, is really where alot of the innovation is going within the cloud, so if you're looking to takeadvantage.

07:29 of some of theinnovation that's happening in the cloud, you know you doing things likemicroservices and your containerizing everything and things like that probablyone of the best ways to.

07:41 take advantage of thatis to use proven technologies for that type of service and usually the pastservices are there so.

07:49 Those are the thingsthat we really see customers, taking advantage of so not only are they youknow, having maybe.

07:57 Know fewer FT heneeded to once they get it into azure they're also getting some features thatthey did not have in the mainframe, especially in the ability to moreseamlessly scale up and down.


Walter Sweat - Astadia, CTO

08:10 That makes sense, soyou just raised a couple of interesting topics i've got a couple of questionsi'd like to follow up with on on that specifically about you know, usingdifferent technologies i'm.

08:23 From your opinion,what are some of the the technical advantages, then that organizations mightsee or is that what you're really talking about being able to kind of do thingsdifferently, where you leverage things like microservices, are there anyothers, you can think of.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

08:38 Why because there'stwo main ones, the ability and more seamlessly scale and then the second onewould be to.

08:45 take advantage ofthings like microservices where you would be able to then sort of change yourway you build applications, make them more agile.

08:54 be able to implementdifferent technologies and azure fortunately embraces that so and doesn'tnecessarily embrace it in a way that uniquely Microsoft so let's say I wantedto do something.

09:08 With open shift well Icould do that on azure to open shifts available in our marketplace.

09:14 And by the way, Itechnologies that steady work with can take advantage of open shift, so I meanall those things are.

09:21 available so there'sthat plus the other one I can rethink how i'm actually say deploying some of myservices within the.

09:31 cloud and i'll giveyou a really good example of that let's say i'm doing you know.

09:37 People say they wantto get around you know batch they want everything to be more real time and oran event driven or something of that nature, well, I think event driven isgreat, but some events you know kind of trigger off batch.

09:51 And you know, and Ireally don't see around.

09:54 A way around that whenyou have things like close of market for financial systems and things like thatthere.

10:00 You unless thebusiness changes which the cloud is not going to just change the businessovernight, although it changes it to some degree.

10:07 But it's not going tochange things like you have market open and market close, so I guess the pointis is that, if I have something that's event driven I don't have to think inthe term of a batch run.

10:18 The way I did it withmy traditional cobalt systems if i'm implementing things with some of my pastservices, I could say hey you know what i'm going to.

10:27 I'm going to talkfinancial systems here i'm going to take every account that I either have inthis core banking system.

10:35 or this investmentsystem that i'm running and i'm going to let's say apply dividends to it, oryou know if its investment or i'm going to apply service fees or.

10:44 distribute interest toit if i'm doing that towards all the accounts in it, I could use something likespark maybe to do that in a much more efficient way of using that compute, tothe point where it might be.

10:59 You know advantageousto actually do an export to you know something that spark can run like.

11:06 A no sequel systemlike cosmos db or even into the parquet, which is an like a data like typesystem.

11:14 Then I could runsomething like data bricks, which is the past service for a spark that as yourhas to do that, and what we found is the time savings.

11:24 You know, especiallyfor the large customers, we can cut several hours off their end of day ofprocessing because of the parallelism that can be done, as opposed to.

11:34 You know, having to dosomething bring it all together sort it then distributed all backed out thenfinally reloaded you can maybe take that all and put that into a single stepwith.

11:46 spark so that's whati'm talking about from a point of view of rethinking when you're takingadvantage of the things that are available for you on the cloud.


Walter Sweat - Astadia, CTO

11:56 that's perfect and youjust segue it perfectly into the second question I wanted to ask, as a followup.

12:04 And that's about howcustomers look what experiences you're seeing where customers are actuallyextending the life and value of their legacy application once they move toazure.

12:15 Being able to usethese new technologies do you find that more people just say i'm going to getto azure and save some money.

12:24 Or are people reallystarting to understand these technologies that you just mentioned and startstarting to be able to re envision the way that they have these applications inplace and what can be done with them.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

12:39 Well, it kind of is alittle bit of both I don't want to say it's it's just one or the other but.

12:45 Typically, the initialthought is saving money you know i'll be very straightforward, the initialthought is okay, I could probably save money.

12:52 But if i'm not gettingvalue out of the effort and i'm introducing risk without getting value is thatreally something that I want to undertake because you know I had one time at.

13:03 A Department ofDefense customers say to me okay So what if you can say me $20 million a yearyeah i'm not going to put a billion dollar asset at risk, save.

13:15 You know 20 million ayear, so you have to look at what but they finally did it, they did it becauseof.

13:22 They were getting morevalue out of making that change if it just came over as is it wasn't worth therisk to them.

13:28 But if they weregetting more value out of the system, then it was worth the risk because theywere actually getting something that they were unable to do.

13:36 You know, on the priorsystem, so I would say that you know that is something that it's not just allmoney, money was the initial.

13:45 thought there, but wewere being shut down because it wasn't worth the risk, but when they looked atthe value they could get out of it by doing.

13:53 Some of thesetransformations that would allow them to take advantage of other things thatare in the cloud, all of a sudden that became something that they wanted to doso.

14:03 I would say that youknow I we still see the customers that are strictly money is is the thing thatthey're after you still do see that, and they can get a significant savings bygetting off the mainframe.

14:16 But more and morecustomers are looking at okay i've got this core IP that's in this, especiallyif they've developed the system's themselves i've got core IP.

14:27 Do I really have torewrite that to maintain it do I have to replace that and Oh, by the way,replacing it is not.

14:35 As simple as it sounds,because these applications could be 20 3040 years old and what does that meanthey've done a lot of customizations you know along the way.

14:44 And there's probablynot something that they can get from a vendor that has all of theircustomizations so the ideal case would be.

14:53 That you can save theIP that's critical to your business that is giving you some kind ofdifferentiation, but also be able to take advantage of other systems.

15:03 As they are availableand usually going to the cloud, you can do one or two things when you get theregoing into the cloud I always recommend if you're moving an application.

15:11 really consider movingthat data to don't leave the data on the mainframe there's other other reasonsthan just innovation for that, but.

15:19 let's say that you'vemade that decision once that data is over, on the cloud, you have a lot ofoptions, but the second is the application.

15:27 Okay, when I have theapplication on you know off the mainframe I can now interact with it more readilywith other systems so.

15:35 For instance, a lot ofthe technologies allow interfaces into either Java.

15:41 Applications or.netapplications pretty easily, and you know, I was already talking about using aspark application that was using scala which is kind of like spark version ofJava.

15:53 You know, to run someof those things, so you just get a lot of different possibilities that come up,and you know, as you mentioned all customers are unique to some degree, whenthey get to be especially the big ones.

16:07 So what we have to dois find you know what is real value that we can do for them, you know as partof making this migration and that brings up a good point that you know.

16:19 Microsoft it's notjust about moving you know with my team We work very closely with our accountteams that are also very enterprise focused.

16:29 And quite often theyalready have some other things that are moving into azure perhaps they've got.

16:35 some kind of datasystems with sequel server or even Oracle, for that matter, that they're movinginto azure and they have some systems that are already set up in aninfrastructure in place.

16:46 Well, what we say, isif the applications are being moved into that let's take and incorporate themainframe into your mainstream way of moving forward sort of your data Centerof the future.

17:00 That also extends thelife of the mainframe application now it's part of the mainstream of what thatcustomers taking forward.

17:08 So I would say that'sthe other big thing is, you know don't think of the mainframe is applicationsliving in isolation, when they're brought to the cloud think of them as beingyou know integral part of everything else that the customer is doing within thecloud.


Walter Sweat - Astadia, CTO

17:23We have seen the same thing, wherepeople truly start to see the benefit of not having two distinct environmentsin two distinct systems.

17:32 But, looking for thosepoints where you can benefit by letting them talk to each other and lettingthem share data letting them share access to logic, so I totally agree with youabout that.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

17:44 You guys were great.


Walter Sweat - Astadia, CTO

17:47From your perspective, how do you feelthat people should really think about addressing concerns for performance bothcompute and io and security when they're going through transformations.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

17:59 Well, you know all ofthem need to be addressed, now, you know quite frankly I think computes theeasiest one to to address, I mean you know customers quite often may think wellthis mainframes just big powerful box, well, it is, but.

18:13 If you look at thepower that's in the cloud and really i'd be any clouds vendor but you knowwe'll just concentrate on azure we've got more equivalent of MIPS now we don'tcall NIPs.

18:27 You know what we callthem as your compute units, by the way, so we've got our own thing from NIPsnow.

18:33 But anyway, my pointis is we actually have magnitude of more power with the it's actually in thecloud that you have access to, and that can either be scaling up so.

18:44 we've run there'ssomething called the Z ref benchmark, you might be familiar with, but we've runthat with a lot of our partners where we can easily scale up to.

18:56 You know just for asingle lpr of 25 to 30,000 MIPS now my experience is that very, very fewcustomers, even the 300,000 MIPS customers have a single lpr that using morethan 25 or 30,000 MIPS.

19:13 So you know, even ifyou need to scale up you have that you know compute capability with in azurebut even better is taking advantage of being able to scale out.

19:24 that's where, if we cancontainerized this if we can deploy it in a ks, then we can.

19:31 You know, we one ofthe customers, we worked with we took them from 65,000 NIPs to a solution thatwas running in.

19:41 You know our azureCooper nettie so service aka es where we had you know 20 nodes of.

19:49 For V cpus each andthat was doing what the online was doing with in that you know 65,000 nipsamainframe that they had, and then we then use the.

20:01 cosmos db is the isthe data side, and you know we really easily able to meet the compute theirnet, but that does bring up the next piece, which is what do we do with thedata actually think that that's.

20:14 Really more of a thingthat we spend time upon, because one thing mainframes do really well is movedata around.

20:22 They they they may notbe doing all the fanciest compute in the world, you know they're not.

20:28 I don't think of amainframe is an htc high performance computer system, but you know, some peopledid that with like fortran way way back in the day, like me.

20:38 But uh you know, thesedays, if you're doing you know things with our and stuff like that you'reprobably or you've got it over into.

20:46 you're doingcomputational fluid dynamics or something like that you're probably doing scaleout for that type of compute but the key to being able to have the.

20:54 mainframe do that iois making sure that we got both the I OPS, and the throughput that you need forit, because mainframes are very fast have for that plus a mainframe.

21:08 has to other featuresthat are kind of Nice you've got that a sort of io channel that you've gotwhere they can use something called a hyper sockets to have very, very lowlatency for doing that io we've worked with customers who, even if it's somewhere the.

21:29 You know where wedon't have high latency with the io with if it's moved into azure.

21:35 One customer had10,000 you know round trips to the database for a single you know transactionwell yeah yeah So even if you that only takes.

21:47 One millisecond I justadded 10 milliseconds to my transaction right so so under those conditions.

21:55 You know, you need tobe in the microsecond range right, so we had to get it down to more like to the30 to 50 microseconds Now you can do that.

22:05 What you need to bevery conscious of is the proximity of the solutions when you need that kind ofperformance as well as the overall throughput now as your has something calledproximity pinning.

22:16 That allows you tothen take like the storage and then have that within we call them stamps that'slike the things where the actual physical hardware, is it such that you can.

22:27 move that back andforth very quickly, then also if you really, really need to have a low.

22:35 latency we canintroduce things like HP see technology, we do have the EMS that run an azurethat can take advantage of things I can find a van.

22:45 And then find a band,you know uses something can use something called a DMA.

22:51 Remote to directmemory access which that essentially bypasses the tcp stack that's how youshave from say like.

22:59 One millisecond downinto microseconds so not every customer needs that but, if you do we canarchitected in a manner, where.

23:08 You know youdefinitely can have that and then the last topic is you know, particularly ofinterest these days.

23:15 With the you know allthe you know things out there, like solar, wind and all this all that kind ofthing that happened recently, is making sure your systems are adequately protectedand that's where security is not just about making sure that we've got the.

23:34 You know theauthorization being done, you know the authentication and authorization beingdone it's also about.

23:40 locking things down orhardening things is probably a better word to say everything's locked youprobably can't do anything but anyway.

23:48 Making sure that it'shardened in a manner, so you know what I do is I always ask my customers to.

23:55 You know, consideralways always consider that you know any system is going to have intrusions youjust have to develop it that way, that has to be the mindset.

24:08 of how it's deployed,and if you take that mindset, even if you do have some kind of intrusion inyour system, then you're able to do it, and the other thing is.

24:19 If you do require, forinstance, federal government.

24:23 We do have specialversions of azure which isolate you even more like there's even one that'scalled air gap essentially what that means it's connected that nothing so.

24:37 You know, so that thatis probably the most secure for that you know the highest amount of security,but there's also you know their secret and top secret and then there's regularazure also and what we do is we go through the.

24:55 Even with commercialazure we take, and we.

25:00 have gone through whatthey call the provisional um you know authority to operate PA to for.

25:08 You know systems, suchthat we can you know meet some of the standards, you know there's standards forsort of low medium and high for.

25:18 You know, security andeven commercial as your can meet the high again it's a PA to because it'salways provisional because you have to then prove your specific implementationof it meets the.

25:32requirements, but you know as long asyou do, that other part that I was mentioning making sure that it's hard and aspart of doing that.

25:40 Then you know you canhave a pretty secure environment so like I say it's a little bit of a journey,you have to think about it from that perspective.

25:49 mainframes were sortof fortunate and, but they were a little bit more isolated, but one of thethings you want to take advantage with in the cloud is the fact you're not asisolated.

25:58 Well, if you're doingthat, then you have to make sure that you have a hardened such that you arestill having the same that same degree of.

26:06 Security so as, morespecifically, most of my customers who are in financial services, for instancelet's see if i'm doing access to a financial services system that has you know.

26:20 Personal identifiabledata, you know pii data there's really three steps that are gone through withthat is the minimum this is even before we get into extra hardening is numberone having that.

26:33 You knowauthentication and authorization done properly.

26:37 But the second thingis i'm going to authorize you know also and authenticate also the devices thatthey're coming from it's not just the people it's the devices so i'm going toexpect a certificate from those.

26:52 devices and if youdon't have a certificate that I recognized then i'm going to refuse thatconnection, you know before you even get to say who you are right.

27:02 So you know you soyou're going to have certificates and you know set up so that you can have youknow pls transport layer security setup even before I do anything else.

27:14 And then the lastthing is that this is where we can you know work with you know the end to endencryption.

27:22 of you know, thesystems, so if my stuff is in the end encrypted, even if I have people who aregetting access to it they're not going to do much with it they're going to seea lot of encrypted.

27:35 data, but it has to bein the end I don't know if you will remember, a few years ago, one of the.

27:41 You know, big boxstores had had an issue with one of some of their systems well without goinginto too much, the issue was the end device, it was actually the end device inthe store that was you know the.


Walter Sweat - Astadia, CTO

27:54 unsecured it was weaklink.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

27:56 yeah it was the weaklink it wasn't the main systems so that just shows the importance of going inthe end, with all these things and and there's one other thing that I wouldsay.

28:08 Is that a number ofthe customers, we work with technologies that we have that are also availablethrough third parties.

28:15 That will you knowobfuscate the actual pii data, so what you're seeing has gone through atransformation, you know, by the time it even gets off of where it's.

28:27 You know, you knowstored in an encrypted manner so that's kind of a long winded answer, but thoseare the different steps that we take to make sure that we have.

28:36 You know the type ofsecurity that you need in Microsoft we call that security in depth, because.

28:42 You know it's not justany particular thing that we're depending on as the, the only way that you canget in its every piece along the way, has to be hard and such that.

28:52 You know if you're notauthorized to do it it's a mindset way back when we used to keep the bad peopleout, you know we would program things to keep the bad people out the mindsetnow is only left the good people in that that's that's a completely.


Walter Sweat - Astadia, CTO

29:0 7different way ofparadigm yeah yeah absolutely different way of looking at things so that's whatwe do to make that happen.

29:13 make sense of like aquestion that I always find intriguing I would really like to know yourviewpoint um.

29:21 Do you feel that thereare mainframe technologies that just really don't fit in the cloud are there,people who are doing something so unique that the cloud might not be an optionfor them.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

29:33 Well there's two typesthat don't fit really well in the cloud one of them would be, I have no sourcecode for it okay and.

29:43 And that does happen,and I don't have a solution, there are some vendors that can take load modulescross bought, they have to have the underlying.

29:53 systems that supportthose load module so in order for that to happen, for instance, if the loadmodule You see, I see so they have to have something that would implement cic scalls so.

30:04 If you've got systemsthat are using a third party system, you know that is not directly like youknow what the gift from IBM like I am master cic s and you don't have source,those are really difficult.

30:20 yeah those are toughnow if you're looking at things that maybe you don't want to move on one of theones that i've seen people not move is that.

30:31 they've got somethingthat they've got a lot of integration with third parties already with, and oneof them would be like encryption.

30:40I have a number of customers who haveyou know there's specific hardware, you can get for mainframe that does youknow the encryption.

30:48 chips and devices thatyou can have some customers have that already in place and it's just it makesmore sense, just to use that as sort of the gateway into their.

31:01 system and they justthey just leave the mainframe there for that one reason, and then they move allthe applications, because you still need to have sort of that that that's inessence they're.

31:12 encrypted firewall, ifyou will, to get into their application so you know I see those is difficult tomove, and then the last one is you know, remember, I talked about what's thevalue.

31:25 of moving some ofthese systems, there are some systems that i'm not going to get a whole lot ofvalue, you know from moving them in and is it really worth the risk to movethat and in, in particular, we have some systems that i've run into withgovernments were.

31:46 Without getting intodetails that I can't share it just doesn't make sense to disturb those things,and so we'll just keep them and let them be sunset as opposed to trying to movethem.


Walter Sweat - Astadia, CTO

32:00 So those are the riskreward.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

32:02 Right yeah it's it's awhole risk reward so now if you're if you're going into other things which.

32:09 it's more of a youknow the difficulty if i've got something that was written entirely inassembler.

32:16 Okay there therearen't a lot of good solutions there's some solutions, but it may or may not begood for moving that assembler because it's extremely efficient and.

32:26 Even the systems thatyou know work if you're going to move it across there's some vendors, who havethem they're not going to be as performant as the assembler.

32:34 On the mainframe somaybe those you might want to rethink if that's a you know a good thing to do,and then the last one my favorite is Z tps if you've gotten ctv tpm systems.

32:48 You might want tothink about you know how am I going to re architect those as opposed to.

32:54 How am I going to movethe applications more or less as he is now that's all primarily assembler basedto, but it could also be in C or c++ So those are probably the ones that i'vehit that.

33:07 You know, we have toreally think you know get what's the best way to do it, and maybe we're notgoing to move the applications just the way they are.

33:15 Across we're going tohave to do something in the process for being able to do it to run in theenvironment when you bring it so those are probably the main ones that I canthink of off the top of my head.


Walter Sweat - Astadia, CTO

33:26 Okay perfect I wouldadd the ones that I encountered that are always a little challenging if someonehas 3040 years ago written their own version of ci CS or Ms DC or they'vecreated their own database architecture, those are some unique challenges, notto.

33:43say they can't be moved, but theythey're ones that always kind of make a scratch our head was about the best wayto handle it.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

33:51 yeah I had a customerat a via the BSC application that they wrote they sort of wrote their ownhosting for VSE for their customers and in the process of doing that they havea really nice dsc multi tenant system.

34:07 But it's just like yousaid you know the effort to do that, you know, is really, really hard.


Walter Sweat - Astadia, CTO

34:14 Absolutely, so thereany parting words of wisdom for people who are listening in who are interestedin exploring their cloud journey from the mainframe.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

34:23 Well, I think that umyou know i'm going to go back to what, what do you want to do when you land onazure because a lot of customers.

34:31 You know, have issuesbeyond just the fact that they want to get rid of the mainframe maybe they'vegot applications were there it's hard to replace the programmers like ifthey're in cobalt, or something of that nature.

34:44 And you know they looklike they really do want to refactor those applications, maybe they would liketo change the languages is moving forward so we actually seeing a lot of thatis.

34:56 You know it's not justmoving things over, is as easy as it used to be like maybe 80% moving thecobalt, for instance, as is now we're seeing it more like 6040 maybe 60% stillthe cobalt, but another 40% that.

35:12 want to refactor it ispart of moving it because, again that's where they get the value and from myunderstanding there's some news the stadia has about being able to sort ofexpand capabilities in that manner.


Walter Sweat - Astadia, CTO

35:24 yeah so we're reallyexcited, thank you for mentioning that, with our recent acquisition of n youbacks our goal at sad is to recognize that.

35:33 You know, as we talkedabout before No two environments, are the same know to pain points are thesame, so we want to provide solutions, no matter what people are looking to do.

35:43 And by acquiring andnew backs, we now have the ability to help refactored cobalt and cic isapplications ID Ms applications natural a database to the cloud.

35:56 Of utilizing differentlanguages, so we can go from assembler to a different language, we can take coball and go to C sharp or Java and it just.

36:07 expands the portfolioand, as you know, from having worked with so many organizations as potentialclients over the years it's important to have that flexibility so yeah reallyexciting times for us here right now.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

36:20 And I just add i'veworked with any pets in the past and we've always found them to be a goodpartner for what I would call the refactor type technologies and moving acrossyou know to cloud systems.


Walter Sweat - Astadia, CTO

36:33 Absolutely, well, itlooks like we've come to the end of our time, I really want to thank you fortaking the time for sharing your thought leadership and experiences.

36:43 I know that this willprobably raise many more questions as people kind of digest everything that youshared and I hope that you know as they come up with these questions that theywould.

36:53 You know, want toreach out to you, to be able to kind of learn more about what azure can do forthem.


Larry Mead- Microsoft, Principal ProgramManager, Azure Global

36:59 Okay, no I’m happy I’mactually I’ve got a really simple email it's Larry me la Roi me at MicrosoftCOM, so if anyone has any questions follow up be happy to have him reach outand try to answer them as best as possible it's been my pleasure to have this discussionwith you, Walter.


Walter Sweat - Astadia, CTO

37:18  Larry it's always a treat to get a chance to talk with you, I suredo appreciate it, and for everybody in the audience again, thank you for takingyour time out of your busy day to join us.

37:28 Please visit us at www.astadia.com to check in for upcoming podcasts of shows thatwe have planned we have quite a few plan for heading through the summer, and Ihope you will join us for them, thank you all and have a great rest of yourday.