Jonathan Frost and Jim Dugan, Microsoft

Walter Sweat - Astadia, CTO

00:04 Hi everyone, welcome to the latest edition of Walter's World the podcast series from Astadia.

00:10 For those of you who may not know a stadium is a worldwide systems integrator and consulting company focused on mainframe migration and modernization.

00:19 And this podcast series is really designed to help organizations who are looking at alternatives to what they do on the mainframe today by bringing in industry leaders to help talk about the information that we feel would be really helpful and pertinent for people who are considering alternatives.

00:37 Today, I am especially delighted to have two of my friends from Microsoft, Jim Dugan and Jonathan Frost, both Principal Program Managers at Microsoft here today to talk to you. So Jim and Jonathan, thank you both very much for taking the time to be with us today.

Jim Dugan - Microsoft, Principal Program Manager

00:54 Thanks for having us.

Jonathon Frost - Microsoft, Principal Program Manager

00:56 Thanks Walter, great to be here.

Walter Sweat - Astadia, CTO

00:58 We're looking forward to it, since I had the chance to talk to both of you just kind of if it's okay to kind of bounce questions back and forth and please if Iask a question of one and the other person thinks there's something that's helpful, let's just make this as conversational as possible.

01:15 Jonathan why don't I start with you um you know, one of the things that I think will be really interesting for the audience is to understand that.

01:24 With the project that we recently completed at the US Air Force extremely critical mainframe migration, there were some great lessons learned and information that I think would especially be helpful to organizations in the federal space as they look at.

01:42 You know what they could do with their mainframe environments so before we start with that, though I just like to know what is it that you and Jim do at Microsoft.

Jonathon Frost - Microsoft, Principal Program Manager

01:54 Thanks Walter, so yeah so both Jim and I are part of what we call the the Azure Global Engineering group and in this group we are very federal focused, US federal.

02:05 We also focus on international and commercial as well, in terms of helping our customers who are both in the government and non-government move their legacy, what we call legacy workloads which includes mainframes mid range and other non x86systems into Azure. How we do that is through helping provide pathways education and things like reference architectures and other guidance patterns.

02:34 To help move them into Azure we also help from the end to end and so helping with looking at the early days of trying to figure out the right pathway to.

02:45 When we actually have a delivery partner like yourself Astadia to be involved in that process and help with technical guidance and coaching along the way, all the way through the through the deployments.

02:58 And, as I mentioned about the reference architectures that's another key part of what we do is learning from these experiences and then documenting them and she reference architectures which we have been publishing a number of them, including one that was a junior sized version of what we did for the the US Air Force project.

03:18 On the Azure architecture center so we actually have a mainframe and mid range category on on the Center and the architecture is is one of about 12 that we have published now and we have more on the way.

03:30 We also produce White Papers and we produce other artifacts blog articles other guidance.

03:37 But also being part of engineering is that we like to give engineering feedback and our goal is to make azure the best place for legacy artifacts moving from on Prem into the cloud.

03:50 And so, through our work with, that is, we work with our other engineering cohorts whether it's as your networking as your data as your storage and try to give feedback and improvements that can can we find that with some of these legacy artifacts that there are times, where we need to improve or change azure to to meet the needs of our customers and that's another key part of what we do is make sure the best place possible.

Walter Sweat - Astadia, CTO

04:17 And certainly makes sense, and I really commend you and the Microsoft team for providing that kind of information for an organization that hasn't been through moves, such as this.

04:26 Trying to have as much information available as possible is obviously critical for them so great job on that.

04:34 Thanks Jim, let me turn to you for a second your perspective on mainframe modernization within the Federal Department of Defense and civilian agencies, specifically.

Jim Dugan - Microsoft, Principal Program Manager

04:45Okay sure, thank you for the question is it's a time of incredible opportunities on both the civilian side, and also on the dsp side we've got.

04:58Like for the air force that the air force was facing a huge cost of running this application was hosted, and so they were facing a tell you know.

05:11You know, like 24 to $26 million.

05:14is how much it would have cost to.

05:16host this application, which is crazy, and so they actually wanted to save costs and that and and you know, have the same capabilities, they couldn't.

05:30Change change a lot with it, but, but the huge cost savings that came there, and also the the more capable software that's out there is the ability to run co ball on a modern platform and for it to scale and for.

05:48You not to have to change very much and it also was, I know that at least in the air force side, it took I think it took five or five months before the payback period.

06:00It was a.

06:02project and that included the cloud cost for a year it's it, it was absolutely great and then the ability for these applications once they're moved over for them to.

06:17In on the mainframe side is you have you're limited by the box itself and by moving over to the cloud you're limited by what the actual cloud can give you and you, you choose what the scale is.

06:32And you're again not having to pay for what you're not using but you also, as you start moving more are using more and you need more scale it's just the scale just keeps on going.

Walter Sweat - Astadia, CTO

06:45it's just there right.

Jim Dugan - Microsoft, Principal Program Manager

06:47The other the other key thing with this is that you know that's on the DVD side.

06:52But on the civilian side is.

06:55there's a huge push now with the tms the technology modernization fun that's out there, where civilian agencies can go out and go to you know.

07:07go through the process but, basically, is it gives funding they they keep giving you know, right now, there was.

07:15You know billion dollars just given them through one of the relief funds this technology modernization fun where civilian agencies can go out and ask for money.

07:24And they they can actually fund modernization initiatives, so the time is is great it's it's on tf cio.gov is the site.

07:35To just learn about the process and that's for modernizing your applications moving your applications along and it's that's why i'm saying.

07:44Is it's a great time to look at this you got the tools you got the cloud which scalability and potentially have some funding that's out there, and you also have other examples, within the audience.

Walter Sweat - Astadia, CTO

07:57Oh absolutely you know just having the flexibility to not just save money, but to be able to leverage new and different ways of doing things.

08:07it's it's an amazing time for people now who are considering alternatives to the mainframe and I think that air force project is a real real world example of that Jim let me come back to you with talking jack the Air Force the IMDS migration.

08:24were some of the key benefits of azure as in supporting this warfighter environment and for those who don't know talk about mission critical environments, this is for handling the.

08:37Maintenance for all the wartime aircraft, the icbm being able to take advantage of newer architectures and saving the money for some of the most critical application environments out there.

Jim Dugan - Microsoft, Principal Program Manager

08:53will be at beyond the costing there was additional capabilities that they cannot take advantage of a good example is utilizing.

09:03artificial intelligence machine learning on the data so part of part of the data that they have it's logistics information part data repair data for any of the airframes within the air force.

09:16Well now, you can implement things that we've been doing for several years.

09:23Around predictive maintenance that this this part breaks over here, then you should also replace this part over here because we've seen in the past that they break in pairs so instead of just just doing one.

09:36You want to do both of them so it's actually allowing you to to predict predict, you know when things are going to go bad what you know how.

09:44Based on your workload and have that constantly adjust it and that's just one small example but opening the be really taking the database that was.

09:56Really siloed inside the mainframe putting it into a cloud database opens up a lot of those opportunities and the air force can take advantage of that also.

10:11The benefits of multi you know fault tolerance disaster recovery.

10:16The ability to.

10:20Take the environment, because if you have the environment and you want to do another Dev test environment you you kind of need to have a fast way of.

10:32duplicating that environment, and now they can duplicate the environment of the production system just in in a couple minutes were before.

10:44They number one we're limited by the number of environments that they could use because they were limited by the mainframe but now.

10:50They can actually copy the environment with a single script run it and have it running in parallel.

10:58To control for continual development in that buyer so there's there's tons of other I mean we could probably just list those after the hour and those but I kind of wanted to give you two minutes.

Walter Sweat - Astadia, CTO

11:11And that's really quick, I think that you know we talked about freeing the data and what that really allows people to do.

11:17Rather than being in siloed environment where you only have one way to look at data being able to leverage the technologies that you talked about.

11:27For the air force that really put them in position to do things differently than they've ever been able to do before so as you say, not just cost savings, but also a new way to do business.

11:42So.

Jim Dugan - Microsoft, Principal Program Manager

11:42That that also brings up something that's kind of ties back to the first question you brought up as to what our team does is.

11:51we're also we by getting involved in the opportunities by getting involved in the migrations we're learning things too.

11:59And so we're talking with the other teams, like the Ai te sore or the you know the machine learning team to say hey you know we we.

12:10This customer has the data they're sitting here and we want to make this easier and more prolific for a legacy type customer and that's that's not a thing that we did that underscore from that first question.

Walter Sweat - Astadia, CTO

12:23that's a great point yeah i'm a question that I get asked often is i'm sure you do, and I think it's an important one, so i'm going to ask both of you this one, if you don't mind as.

12:35As federal organizations look to move to azure how secure can they really feel that they'll achieve the same security levels that they see on the mainframe.

Jim Dugan - Microsoft, Principal Program Manager

12:48yeah so so Microsoft has been focused on strong security and privacy protections for for over 10 years we have the the big privacy initiatives and and the trustworthy computing.

13:03That that went through Microsoft, and so it was really kind of burnt into our DNA about everything that we do is we we.

13:12You know, we look security first we look at the differentiators are the privacy differentiators that we have inside the products themselves and we're always looking to expand those and we are there's constant.

13:25You know, if you look at the changes that azure is gone through over several years is it's done done two things it's it's also added capabilities to the products but we've also gone after additional.

13:45capabilities that the government cloud, you know the secret cloud going after you know different things like fed ramp God certifications i'll level sieges.

13:58hipaa you know and support for ios 1075 is among there's tons of others there's there's a lot of different compliance information, I mean.

14:08Compliance that we're following, and that is as caused you know, in the past has caused some changes but.

14:16We go, you know even further than that, by enabling some some different things where you didn't have in the mainframe like the ability to audit.

14:26The ability to to raise alerts, or even to tie in some level of let's say intrusion detection, you might you know you you as you're constantly watching and looking for signs of.

14:42Some sort of intrusion to notify somebody and that level of intrusion could change, based on the activity of the data it could be.

14:53You know, different different variations of you know, the attack vectors change all the time, the cloud is constantly watching that and adjusting to those.

15:03Whereas you don't have that same capability on the mainframe you, you can add intrusion detection, you can add this but you're constantly having to keep up.

15:12With what it's doing dealing with and taking advantage of things that were done in the cloud and the ability for the cloud to learn from those and by by having the system.

15:26In the cloud you you kind of get that you get the you know the ability for the intrusion detection for the for the different levels of auditing for an anomaly detection of the activities that are there.

Jonathon Frost - Microsoft, Principal Program Manager

15:39And just just to build on that to to to make some really great points here is that and something that I think you touched upon just a moment ago, Jim is that.

15:48How will your on premises and you're automating from system, you are in a very slow to change monolithic system and so as the world changes and as intrusion methods change as security models need to become more advanced.

16:05In the cloud are able to be more agile and have better intrusion detection learn.

16:10Better patterns bring in better capabilities, with our security with our privacy and also with our certification levels that you mentioned as well, where on premises, yes, there.

16:22There is security, however, it is very slow to to adjust and adapt where we're able to do things much more quickly and much more agile once the workloads are in our in azure and then the cloud.

Unknown Speaker

16:35Great point.

Walter Sweat - Astadia, CTO

16:37Of Jonathan we talked about security can we talk about the same thing about what people really can expect in terms of scalability and performance, because obviously that is equally as important for people coming from a mainframe environment.

Jonathon Frost - Microsoft, Principal Program Manager

16:52yeah no that's a great point that's a great actually segue from that that concept of being agile and moving to the cloud.

16:59Is that there's so many more options, now that we have and Jim touched upon this earlier in terms of vitamin D, yes, and how we're able to take advantage of that that concept.

17:08So there's everything from scalable past services were simply by using the notion of a slider we can change the amount of capacity, storage and compute.

17:19On workloads as needed and, as well as their capacity in terms of different architectures so to graze apples are azure sequel database hyper scale architecture which has a skill out.

17:31Type architecture that allows us to have more storage in more compute and do fast processing in the architecture.

17:39As well as also on the Open Source side we have on the postgres database, we have a pass service, which is our azure postgres hyper scale which gives us horizontal.

17:51scale out for a postgres database environment.

17:54And they're similar in the name is hyper scale there's there's similar concepts behind them to a little bit different architecture, but what they both do is provide scalability.

18:03And in for both storage in in compute on these platforms and so there's by having those platforms, as well as other.

18:13layers as well as on the application and compute side we have things like aka as as communities services where.

18:21You give a piece of code piece of Java piece of C sharp that's running in a container on these.

18:28Communities clusters and you can have a defined where upfront you might know what your capacity and throughput is in you can set that scale.

18:36Or that we even have auto scaling options, where depending on if it hits a certain threshold.

18:42That you can have more nodes automatically instantiated and then writing that code running those containers on the fly and so there's there's a lot of really fascinating options that we have, and those are just some some some initial examples there's many more as well.

Walter Sweat - Astadia, CTO

18:58it's amazing being an old guy I can feel safe saying it's not your grandfather's compute model anymore, the flexibility that exists is just astounding in azure.

19:10i'm Jonathan i'd like to ask your opinion, how beneficial was it for the air force moving he mentioned this before, to a pez database service and in terms of being able to expand what they could do with Ai and machine learning.

Jonathon Frost - Microsoft, Principal Program Manager

19:26yeah that's that's a really great point and.

19:29So really in Jim hit upon this earlier as well, but i'll build on on what he said is that.

19:35When the key things is that you know the data is now in azure sequel database and previously it was on you know we're on a.

19:43Different system and high article database architecture and structure and by being in a relational database.

19:51In in azure on a has database, we have a lot more opportunities, where the data is much more accessible to other services of course with security and with private endpoints and all the right.

20:03Regulations and security that that should go into that is in place.

20:08But what's now an option to us, especially when it comes to ml is that this data is now in a very good position to be used as training data.

20:17For for ml models and an azure we have a very nice selection of of ml possibilities so everything from our azure ml.

20:29studio in toolkit where you can write your own model and then use that data for training and create your own models into your own deployments.

20:39As well as we have what we call the azure kinds of services which are actually pre baked ml Ai models, where the Everything is already developed in simple university to provide it your your training data and as well as your input data, and it is a preset.

21:03components that you're able to to leverage and so Those are some of the options are there, and these are all.

21:09themselves as your services that can be easily plugged in to data now being in a in the past database.

Walter Sweat - Astadia, CTO

21:16make sense perfect Thank you Kim i'm going to switch back over.

21:20To us, you don't mind um how important do you feel it is to have minimal change to the legacy applications and the source code when migrating for federal agencies when they're moving to azure.

Jim Dugan - Microsoft, Principal Program Manager

21:35It i'll give the perfect consulting answer is it's you know it's in this in the air force, it was very important, because number one is they, they couldn't retrain everybody.

21:49You got to realize that having a worldwide deployment of aircraft and management and repair of those aircraft they they don't really have time to train somebody on a new system.

22:04They that they always have to be running and so, in that case and.

22:11You know you don't really want the screens change okay fine the screens aren't you know the screens didn't change when you know, at least on this particular application.

22:19But they've also previously at least the air force previously went through a rewrite effort that didn't work they they tried to rewrite the application, they spent a lot of money doing that it didn't work, so they really didn't want to change the code.

22:37So in it with the air force was very important not to have the code change, and so they were used to maintaining the cobalt code.

22:46They they they didn't want it to change, so it was important you've got other customers that have different reliance on on the source code that may want to.

22:58Their their system might be changing over the where they're doing Java C sharp or any other you know, a a.net language.

23:10that they want the code actually change and there's there's tools and capabilities for really transforming that, if need be, if that's what the customer wants to do.

23:21So again, it comes down to it depends the code doesn't have to change, and this case.

23:26There was business rules that they had that that that they couldn't change a line of code they didn't want to change anything, and so.

23:33From that it was it was very important, but you know they're they're you're you're kind of having to look at it's like what, what are we trying to move towards and they didn't want to have to have to bring in a bunch of new developers they have you know.

23:53Probably millions of years of talent of cobalt in the code itself having looked You know yourself also having looked at it's like it was it was a lot of cobalt.

24:03That that made its way across, and so they they did you know, so I guess it comes down to it depends on the on the customers requirements bit.

24:13And there's any old is okay.

Jonathon Frost - Microsoft, Principal Program Manager

24:15To say, and I mean just to build on that, I mean Jen is a really good point I mean it does dependent I think that's where you know.

24:23upfront figuring out the pathway is a real host or refactor or re architecture kind of three buckets we kind of see things and it's kind of our point of view and.

24:33In this case it re host made a lot of sense both business and from technology.

24:40And so it worked out really well there's lessons learned previously that we were able to to leverage that that lean towards rehouse but also looking at the full end to end as Jim said rehearse my sense.

24:51was nice as though is that just because it's a rehearsed it still runs in azure and we're able to leverage, a lot of has services, the database as your data factory.

25:01azure blob storage has a lot of things that were around the host environment that even though is cobalt coastal running without a line of code change.

25:09there's also a lot of modern services there we're also running at the same time, so it's kind of a hybrid.

25:13In some ways, and so that's so just because it's a host in codes and changing there's still a lot of opportunity to use modern as your services, and that was on this case in this order with other customers as well.

Walter Sweat - Astadia, CTO

25:25And it sort of seems that you know it's not necessarily a one and done if an organization wants to do the classic we host.

25:32Take those savings and reinvest them and then come back and change to a more modern language they can it's part of a pathway, as you described, there is no one exact landing point for everybody, so, providing that flexibility is important.

25:47Jonathan i'd like to ask you something about disaster recovery.

25:51How important and what are some of the benefits of monitoring and disaster recovery when the application to be moved to the cloud.

Jonathon Frost - Microsoft, Principal Program Manager

25:59yeah no that's that's a.

26:02really important question because that's when that comes up a lot and that's it's something where.

26:07These legacy systems and mainframes they do have a level of redundancy built into them on premises, however, they don't always have quite what the customer really does it's necessarily actually really, really want.

26:20And so we found that by moving these workloads into azure we can actually bring you to the abilities in terms of disaster recovery geographic redundancy.

26:29An auto fail over that are net new capabilities that make when disasters happen when issues occur, it can actually be a more streamlined process and bring things that were impossible so is a very important thing is always a key early conversation.

26:46That we have in terms of a technical factor in terms of going to the cloud and the capability to enlarge, this is a common one, where this is one where it's a.

26:58Something that we're able to to to easily and.

27:03Without a lot of complexity, to be able to bring to the table for that capability so so on those on those fronts and then really it's it's the notion of you know auto fail or groups, which is a key feature in azure sequel.

27:15database, the database, where you can have in one region, your database running and then, in the case of there is either a failure of some sort.

27:27of various other database level, you can have an automatic fail over to another copy and another asher region and the same endpoint there'll be the same.

27:36endpoint that you use so that way your application layer doesn't have to change it will simply redirect to that new.

27:44That new environment with it with it with a replica of the data.

27:48So there are things like that that are unlocked with moving into azure and then beyond that is also the geo replication so there's various services there's there's ash database there's azure blob storage and other ones where the data is.

28:00Not only replicated multiple times within the azure region, but then you can choose to have it replicated outside the region into another azure region and it gives you this really great level of redundancy and so that way you have that and one more.

28:14Quick dimension Now this is with with I MDS was one of the other key capabilities.

28:20That was unlocked was the ability to do a fast restores of databases, so be able to spin up a new environment do a restore and and have that environment available was a very key thing and so.

28:33This will be where i'm premises to in order to provision, a new environment new LP or or the mainframe to to.

28:40You know replicate data all that can is time is expensive, it takes cost and MIPS and all these aspects were in the cloud it's it's faster, and it can also be cost savings as well.

Walter Sweat - Astadia, CTO

28:54into the number of customers companies i've talked to over the years, who say we've never really had a successful Dr test for whatever reason, we ran out of.

29:03time we ran out of money, we ran out of resorts, and they never really proved that it would work, then when they go to azure they're amazed at how easy it can be.

Jonathon Frost - Microsoft, Principal Program Manager

29:14Absolutely.

Walter Sweat - Astadia, CTO

29:15A jump on that to jump back to something you mentioned before about devops.

29:20From your perspective, what are the benefits to implementing devops and agile methodologies, especially for federal agencies.

Jim Dugan - Microsoft, Principal Program Manager

29:29Well, historically, you know federal agencies kind of followed waterfall approach and there's a lot of project management that has gone into into kind of trying to move to more of an agile type devops approach, one of the you know, a couple things that was.

29:52I guess an impediment to that was the system's themselves and a couple of you know, a couple other.

30:00items that that kind of but but by moving into the cloud enabled at least this you know, in the air force.

30:09Where they moved from model where they're now trying to do, continuous integration continuous delivery they're trying to fail fast, just like one thing that Jonathan.

30:19Just brought up a second ago about the ability to duplicate the system, so if you need to be able to take what is in production run a script have it running over you might.

30:34You know, lower the amount of for for testing but say okay let's take a picture of production as of right this minute.

30:43And let's do a snapshot copy over here and then we'll run against that snapshot copy and will only give it two cpus and we'll throw a lot of.

30:54workload or something like that, so it allows them to fail fast and start adopting some of these more agile practices, where they they can now divide things into smaller teams, they can.

31:09They can they can now you know go through and start you know hammering different things out dividing workload.

31:18Because before they they had to coordinate when the workload, or when they're testing was being done, and now these individual teams.

31:26have their own environment that they can base off the production code and then also it enables some some different tools like devops and things like that that that are part of the cloud where.

31:40You can check in your changes, and you, you have full access to tools such as you know, deep buggers and things like that that you really were kind of limited in your capabilities previously.

31:58yeah there's tons more.

Jonathon Frost - Microsoft, Principal Program Manager

32:00yeah the bar yeah exactly and actually just just just just add a few more mean to jim's point about the the small teams and the.

32:09ability to to do improvements is branch management and so by having a modern devops environment and be able to have access to the modern source control.

32:19Is that be able to have a modern branching strategy where you can have your main production.

32:25branch will also be doing your improvements to testing try out things and other branches and have your own azure separate environment for those branches.

32:36Do your experimentation hammer things out and then, when you're ready merge that back into your into your main line.

32:42These all these things are now possible in azure Dev OPS and Dev OPS in general is is a killer for doing that, but then, of course, using tools like agile devops visual studio and other tools like that brings all that together into a nice nice nice tool set.

Jim Dugan - Microsoft, Principal Program Manager

32:56So the automated also he automated deployment.

Jonathon Frost - Microsoft, Principal Program Manager

32:59The part of me yeah.

33:00My mentor absolutely they.

Jim Dugan - Microsoft, Principal Program Manager

33:03You know they didn't have access to those previously and then I think we're showing that to them, they were like Oh, this is great I wish I would have had that.

Walter Sweat - Astadia, CTO

33:12Absolutely.

Jim Dugan - Microsoft, Principal Program Manager

33:14I think the buggers.

Walter Sweat - Astadia, CTO

33:15yeah from the from the MDS perspective.

33:19You know, obviously, was a very successful project, but there were lessons learned, there were things that had to be accounted for, I think this is going to be really important to our listening audience.

33:29And i'd like to ask this to both of you from an architectural viewpoint, what are some of the biggest challenges that agencies need to be aware of, as they contemplate think about a migration to azure.

Jim Dugan - Microsoft, Principal Program Manager

33:46I think.

33:47I guess initially would be.

33:52The first thing that kind of jumps to mind is testing is the ability to generate and have tools to generate.

34:02Look at the load testing environment or the ability to generate synthetic load against your environment, so if you have let's say you know.

34:14users and you have the ability for those users to hit the application and you've recorded that and then you can.

34:22spawn off additional clients let's say in the cloud worker clients that would now hammer your system itself and the ability for for running those the ability or the need for that is generating first generating synthetic load and instead of just you know, having.

34:45What you what you think is a load.

34:48You wanted to try and match as closely as possible to what the.

34:55What the production workload looks like so you definitely want to spend as much time as you can initially doing that i'd say that that was one of the one of the bigger challenges is.

35:08What was originally seen as the load test really wasn't a load test and it had to be generated on the fly.

35:16It you know just running reports good example is just running reports over and over and over isn't really good load test on the system it's good for if everything you're running is is is a reports but that's not really the case.

35:31you're running reports and having to balance reports with transactional workload so it's just even more important, I think.

35:39To have the transactional workload because reports anybody can run reports, you can go in and and easily do that the challenge comes in, is when when the users in the transactional volume.

35:52Are mimicking that transactions of the user activity and then being able to run the reports on it so it's it's well we're spending time up front thinking about how you generate the synthetic load and with that synthetic workloads look like our system, because our goal in this.

36:13Another thing is on jonathan's on our side, our goal is to break ash is if we can break azure great because, if we can break it.

36:23Then by throwing additional workload at it by synthetically generating huge amounts of workload, we want to see.

36:31What it looks like when it breaks beforehand, this is before you know before it goes to production or anything like that, just like what we're doing you want to break your application and see what happens when let's say you know you hit the scale limit.

36:47What happens when you, you know you hit the the memory limit or whatever limit that you have, and how does it behave and then you take that.

36:57And now that you know how it behaves you can start putting some things in place to look, and this would be another thing looking up front is if it breaks, what does it look like leading up to that break do we get any warnings and now we can have alerts and we can have all sorts of different.

37:18Pre warnings before we hit a failure so thinking about those and realizing you, you know again it gets back to the synthetic testing is the ability to hammer your application with as big a hammer, as you can hit it with to see what happens when the application.

37:37hits that and then being able to document that knowledge and learn it before it hits it in production, the second thing is, if you do hit.

37:48Any activity like that you want to go back and do kind of root cause analysis and and take a look and see what happens so that you can also.

37:57You might not have tested or you know it's a real world thing like a denial of service attack or you know, whatever that that you were hitting you just want to make sure that that you you're constantly learning and constantly bringing in those messages and.

38:17updating and evolving your alerting so that you're taking these these lessons that I learned and and alerting on them and looking at those those other types of things.

38:31I guess another one would be.

38:34Having an architecture that scales horizontally vertically just where can you scale.

38:42You know, Jonathan and brought up before the hyper scale database architecture.

38:46That can you know there's different levels of scale there there's this was actually a classic kind of three tier application architecture, where can the application.

38:57scale itself, there was a web server that's on the front end the ability to take that and put that into a cloud based web server so that it can dynamically handle the scale if the Web activity goes up or it goes down.

39:13So it's it's kind of the auto scale capabilities that enable some of the scalable architecture.

Walter Sweat - Astadia, CTO

39:20It just seems to me that is the difference in being proactive and reactive, as you say, be prepared to test it be prepared to.

39:29Put stress on it and to measure and monitor what happens so that you're not surprised when something happens so that plan that methodology and approach is really critical as you move forward, I absolutely agree.

Jim Dugan - Microsoft, Principal Program Manager

39:44And one other thing that's kind of that that you want to take into account is you've got the users in our case in the we're in the.

39:54US air force's case users are coming in from everywhere, and so we can throw load, we can do everything in the cloud, but you want to make sure that you pay special attention to the network, that of the data from users, that are coming in, because.

40:11You make good example is air force Well, this is coming in from overseas, what if it's coming in from overseas it's probably going to be hitting a satellite.

40:21So you want to make sure that you're testing a high latency latency environment.

40:26So that you're not all of a sudden hey it's running real fast and in everything in the cloud, and we can handle with the scale, but how are you doing.

40:35With a second and a half latency of a cloud that's coming in, now there are tools that you can use, and you can test for that, but just make sure that you do that, from a from a lessons learned perspective.

Walter Sweat - Astadia, CTO

40:47Absolutely makes sense yeah well guys it looks like we're at the end of our time, I wanted to thank you both for taking the time to be here and if people who are listening more to learn more about what you and your team's does, how should they best reach out to you.

Jim Dugan - Microsoft, Principal Program Manager

41:04Jonathan i'll defer to you on that.

Jonathon Frost - Microsoft, Principal Program Manager

41:06Alright sounds good yeah thanks thanks Jim so there's a few ways to reach out to us one great way is that we have some number of artifacts that are published on the azure architecture Center which there has links to.

41:19an email alias that comes into our routines inbox collectively, so you can go to a k.ms slash ref Arc ar E, F ar ch and so that will go to the reference architecture.

41:37Libraries and from there, you can navigate to the mainframe and mid range category in there you'll see our collection of reference architectures.

41:46And, as well as then those will link off to other white papers in artifacts that we publish this well, then we have a email alias as well, that is.

41:55available on those that if there's interest from partners customers who want to work with us have questions and require help that there will be an email alias there as well, oh, it is a legacy number to azure at Microsoft COM, but it'll be on that reference architecture site.

Walter Sweat - Astadia, CTO

42:15Perfect Jim and Jonathan again thank you both so much for taking the time to join us today, this was an extremely informative and helpful session and i'm sure the people listening have thoroughly enjoyed listening to you off this thanks so much guys.

Jim Dugan - Microsoft, Principal Program Manager

42:29yeah thanks for having us Walter.

Jonathon Frost - Microsoft, Principal Program Manager

42:31Thank you.

Walter Sweat - Astadia, CTO

42:32And for everyone in the audience, thank you for taking the time out of your busy day to join this latest edition of the Walter world podcast series, if you want to see some of the upcoming sessions, that we have planned or, if you want to learn more about the airport project, please take a look at www dot a stadia COM so again everyone, thank you very much, and until next time we'll see you then Thank you.

‍

‍